Creating 3D models has always been challenging and time-consuming, often taking hours or even days of hard work. This difficulty has been a significant roadblock for many aspiring artists and designers looking to break into 3D content creation. However, a promising solution has emerged in the form of Stable 3D from Stability AI, a leading generative AI startup. This innovative tool aims to revolutionize the landscape of 3D modeling, making it more accessible to everyone, regardless of their level of expertise.

Stable 3D offers a simplified approach to 3D modeling that eliminates much of the complexity traditionally associated with the process. It allows users to create draft-quality 3D objects in just a matter of minutes, using either text prompts or images as input. Even if you’re not an experienced 3D artist, you can now utilize the power of AI to generate impressive 3D models quickly and easily.

What sets Stable 3D apart is its impressive capability to produce 3D files in the standard .obj format. These files can be refined and customized using popular 3D software like Blender or Maya. Additionally, Stable 3D seamlessly integrates with popular game engines such as Unreal Engine 5 or Unity, enabling independent artists, designers, and developers to churn out thousands of 3D objects daily. This transformative feature is set to revolutionize the creative processes for countless individuals and open up new possibilities in the world of 3D content creation.

But Stable 3D isn’t the only exciting development coming from Stability AI. The company is also offering other previews, such as Sky Replacer, which allows users to replace skies in images swiftly, and Stable FineTuning, a tool that enables enterprises and developers to fine-tune pictures, objects, and styles with ease. Additionally, Stability AI integrates Content Credentials and invisible watermarking for generated images, ensuring security and authenticity, an important consideration in today’s digital world.

Stability AI has already significantly impacted with its Stable Audio tool, which empowers users to generate original music and sound effects using AI. Whether you’re an amateur or a professional, this user-friendly interface allows you to create a diverse range of musical compositions with ease. This is another testament to Stability AI’s commitment to making advanced creative tools accessible to many users.

In conclusion, Stability AI’s mission to democratize creative content creation is steadily taking shape. Their expanding range of generative AI models, spanning audio, images, language, and 3D, reshapes the creative landscape. The introduction of Stable 3D represents a significant step forward and hints at even more transformative changes. The integration of these various technologies suggests that we may soon witness the emergence of highly capable multimodal AI models, unlocking creative possibilities for a broader audience. The democratization of creative content lies at the core of Stability AI’s vision, and with offerings like Stable 3D, they are closer than ever to making this vision a reality.

Check out the Reference Article. All credit for this research goes to the researchers of this project. Also, don’t forget to join our 32k+ ML SubReddit, 40k+ Facebook Community, Discord Channel, and Email Newsletter, where we share the latest AI research news, cool AI projects, and more.

If you like our work, you will love our newsletter..

We are also on Telegram and WhatsApp.

Weather forecasting stands as a complex and crucial aspect of meteorological research, as accurate predictions of future weather patterns remain a challenging endeavour. With the integration of diverse data sources and the need for high-resolution spatial inputs, the task becomes increasingly intricate. In response to these challenges, recent research, MetNet-3, presents a comprehensive neural network-based model that aims to tackle these complexities. By harnessing a wide array of data inputs, including radar data, satellite imagery, assimilated weather state data, and ground weather station measurements, MetNet-3 strives to generate highly accurate and detailed weather predictions, signifying a significant step forward in meteorological research.

At the forefront of cutting-edge meteorological research, the emergence of MetNet-3 marks a significant breakthrough. Developed by a team of dedicated and innovative researchers, this neural network model represents a holistic approach to weather forecasting. Unlike traditional methods, MetNet-3 seamlessly integrates various data sources, such as radar data, satellite images, assimilated weather state information, and ground weather station reports. This comprehensive integration allows for producing highly detailed and high-resolution weather predictions, heralding a substantial advancement in the field. This novel approach promises to enhance the precision and reliability of weather forecasting models and ultimately benefit various sectors reliant on accurate weather predictions, including agriculture, transportation, and disaster management.

MetNet-3’s methodology is founded on a sophisticated three-part neural network framework, encompassing topographical embeddings, a U-Net backbone, and a modified MaxVit transformer. By implementing topographical embeddings, the model demonstrates the capacity to automatically extract and employ critical topographical data, thereby enhancing its ability to discern crucial spatial patterns and relationships. The incorporation of high-resolution and low-resolution inputs, along with a unique lead time conditioning mechanism, underlines the model’s proficiency in generating accurate weather forecasts, even for extended lead times. Additionally, the innovative use of model parallelism in the hardware configuration optimizes computational efficiency, enabling the model to handle substantial data inputs effectively. This aspect solidifies the potential of MetNet-3 as an essential tool in meteorological research and weather forecasting.

In summary, the development of MetNet-3 represents a significant leap forward in meteorological research. By addressing persistent challenges associated with weather forecasting, the research team has introduced a sophisticated and comprehensive model capable of processing diverse data inputs to produce precise and high-resolution weather predictions. The incorporation of advanced techniques, including topographical embeddings and model parallelism, serves as a testament to the robustness and adaptability of the proposed solution. MetNet-3 presents a promising avenue for enhancing the precision and reliability of weather forecasting models, ultimately facilitating more effective decision-making across various sectors heavily reliant on accurate weather predictions. As a result, this innovative model has the potential to revolutionize the field of meteorological research and contribute significantly to the advancement of weather forecasting technologies worldwide.

Check out the Paper. All credit for this research goes to the researchers of this project. Also, don’t forget to join our 32k+ ML SubReddit, 40k+ Facebook Community, Discord Channel, and Email Newsletter, where we share the latest AI research news, cool AI projects, and more.

If you like our work, you will love our newsletter..

We are also on Telegram and WhatsApp.

Samsung has unveiled a new artificial intelligence (AI) language model called Gauss, which is being touted as a competitor to OpenAI’s ChatGPT. Gauss is a generative pre-trained transformer model that can be used for a variety of tasks, including natural language processing (NLP), machine translation, and text generation.

Key Features of Gauss

Gauss is based on a new AI architecture that Samsung has developed. This architecture is designed to be more efficient and scalable than previous architectures, and it allows Gauss to process information more quickly and accurately.

Gauss is also able to learn from new data more quickly than previous AI models. This is because Gauss is able to identify and exploit patterns in data that other models may not be able to see.

What Gauss Can Do?

Gauss can be used for a variety of tasks, including the following:

Natural language processing (NLP): Gauss can be used to understand and generate human language. This includes tasks such as machine translation, text summarization, and question answering.

Machine translation: Gauss can be used to translate text from one language to another. This includes languages such as English, French, Spanish, Chinese, and Japanese.

Text generation: Gauss can be used to generate new text. This includes tasks such as writing creative content, generating code, and composing music.

Samsung’s Goals for Gauss

Samsung has stated that it hopes that Gauss will be used to develop new and innovative AI-powered products and services. The company believes that Gauss has the potential to revolutionize the way we interact with computers.

Key Takeaways

Samsung has unveiled a new AI language model called Gauss, which is being touted as a competitor to OpenAI’s ChatGPT.

Gauss is a generative pre-trained transformer model that can be used for a variety of tasks, including NLP, machine translation, and text generation.

Gauss is based on a new AI architecture that Samsung has developed.

Gauss is able to learn from new data more quickly than previous AI models.

Gauss can be used for a variety of tasks, including NLP, machine translation, and text generation.

Samsung hopes that Gauss will be used to develop new and innovative AI-powered products and services.

References:

https://me.mashable.com/tech/34602/samsung-unveils-chatgpt-alternative-gauss-heres-what-all-it-can-do#google_vignette

https://www.ithome.com/0/730/847.htm

https://www.hayo.com/article/654aee5ccf78634039cd63d3

In the race for AI supremacy, a Chinese AI start-up, Baichuan Intelligent, has unveiled its latest large language model, the Baichuan2-192K, setting new benchmarks in processing long text prompts. This development highlights China’s determination to establish itself as a frontrunner in the global AI landscape.

The demand for AI models capable of handling large text prompts, such as novels, legal documents, and financial reports, is on the rise. Traditional models often struggle with extended text, and there’s a need for more powerful and efficient solutions in various industries. 

Currently, the AI landscape is dominated by Western giants like OpenAI and Meta, which have been continuously innovating and releasing sophisticated models. Baichuan Intelligent’s new release, the Baichuan2-192K, challenges these established players.

Baichuan Intelligent, founded by Sogou’s founder Wang Xiaochuan, has introduced the Baichuan2-192K, a groundbreaking large language model. This model boasts a remarkable ‘context window,’ enabling it to process approximately 350,000 Chinese characters in one go. In comparison, it surpasses OpenAI’s GPT-4-32k by 14 times and Amazon-backed Anthropic’s Claude 2 by 4.4 times, making it a powerful tool for handling long-form text prompts.

Baichuan2-192K’s key innovation lies in its ability to process extensive text seamlessly. It excels in digesting and summarizing novels, offering quality responses, and understanding long text, as demonstrated by test results from LongEval, a project initiated by the University of California, Berkeley, and other US institutions. The model’s exceptional context length is achieved through technical innovations in dynamic positional encoding and distributed training frameworks without sacrificing performance. Baichuan2-192K’s outstanding capability positions it as an essential tool for businesses in industries such as legal, media, and finance. Its ability to process and generate long text is vital in these sectors. However, it’s important to note that the capacity to process more information does not necessarily make an AI model better than its peers, as highlighted by joint research from Stanford University and UC Berkeley.

Baichuan Intelligent’s rapid rise in the AI sector, including the recent entry into the unicorn club just six months after its founding, demonstrates China’s commitment to pushing the boundaries of AI technology. While American firms currently hold the lead in AI hardware and software, Baichuan’s aggressive strategy and technological innovations showcase the evolving landscape of AI. The unveiling of Baichuan2-192K is evidence that the race for AI supremacy is far from over, with China determined to challenge the dominance of Western giants in the field. Baichuan2-192K is a groundbreaking model that pushes the boundaries of AI technology, particularly in handling long text prompts. Its exceptional context length and quality responses make it a valuable tool for various industries.

All credit for this research goes to the researchers of this project. Also, don’t forget to join our 32k+ ML SubReddit, 40k+ Facebook Community, Discord Channel, and Email Newsletter, where we share the latest AI research news, cool AI projects, and more.

If you like our work, you will love our newsletter..

We are also on Telegram and WhatsApp.

References:

https://www.donews.com/news/detail/1/3749317.html

https://finance.yahoo.com/news/chinese-ai-start-baichuan-claims-093000489.html

https://www.hayo.com/article/653f4e2b0e9394e0e72011db

A team of researchers from Hong Kong introduced two open-source diffusion models for high-quality video generation. The text-to-video (T2V) model generates cinematic-quality videos from text input, surpassing other open-source T2V models in performance. On the other hand, the image-to-video (I2V) model converts a reference image into a video while preserving content, structure, and style. These models are expected to advance video generation technology in academia and industry, providing valuable resources for researchers and engineers.

Diffusion models (DMs) have excelled in content generation, including text-to-image and video generation. Video Diffusion Models (VDMs) like Make-A-Video, Imagen Video, and others have extended the Stable Diffusion (SD) framework to ensure temporal consistency in open-source T2V models. However, these models have resolution, quality, and composition limitations.  These models outperform existing open-source T2V models, advancing technology in the community.

Generative models, particularly diffusion models, have advanced image and video generation. While open-source text-to-image (T2I) models exist, T2V models are limited. T2V includes temporal attention layers and joint training for consistency, while I2V preserves image content and structure. By sharing these models, the researchers aim to bolster the open-source community and push video generation technology forward.

The study presents two diffusion models: T2V and I2V. T2V employs a 3D U-Net architecture with spatial-temporal blocks, convolutional layers, spatial and temporal transformers, and dual cross-attention layers to align text and image embeddings. I2V transforms images into video clips, preserving content, structure, and style. Both models use a learnable projection network for training. Evaluation involves metrics for video quality and alignment between text and video.

The proposed T2V and I2V models excel in video quality and text-video alignment, surpassing other open-source models. T2V employs a denoising 3D U-Net architecture, delivering high visual fidelity in generated videos. I2V effectively transforms images into video clips, preserving content, structure, and style. Comparative analysis against models like Gen-2, Pika Labs, and ModelScope highlights their superior performance in visual quality, text-video alignment, temporal consistency, and motion quality.

In conclusion, the recent introduction of T2V and I2V models for generating videos has shown great potential in advancing technological advancements in the community. While these models have demonstrated superior performance in terms of video quality and text-video alignment, there is still a need for future improvements in areas such as duration, resolution, and motion quality of generated videos. However, with the development of these open-source models, researchers believe that further improvements in this field will be possible.

In the future, one could consider adding frames and creating a frame-interpolation model to extend the model duration beyond 2 seconds. To improve the resolution, collaborating with ScaleCrafter or using spatial upscaling could be explored. It may be advisable to work with higher-quality data to enhance the motion and visual quality. Including image prompts and researching image conditional branches could also be potential areas to explore for creating dynamic content with improved visual fidelity using the diffusion model.

Check out the Paper, Github, and Project. All credit for this research goes to the researchers of this project. Also, don’t forget to join our 32k+ ML SubReddit, 40k+ Facebook Community, Discord Channel, and Email Newsletter, where we share the latest AI research news, cool AI projects, and more.

If you like our work, you will love our newsletter..

We are also on Telegram and WhatsApp.

ChipNeMo explores the utilisation of LLMs for industrial chip design, employing domain adaptation techniques rather than relying on off-the-shelf LLMs. These techniques involve custom tokenisation, domain-adaptive pretraining, supervised fine-tuning with domain-specific guidance, and domain-adapted retrieval models. The study evaluates these methods through three LLM applications in chip design, resulting in notable performance enhancements compared to general-purpose models. It enables substantial model size reduction with equal or improved performance across various design tasks while highlighting the potential for further refinement in domain-adapted LLM approaches.

The study explores domain-specific applications of LLMs in chip design, emphasising the presence of proprietary data in various domains. It delves into retrieval augmented generation to enhance knowledge-intensive NLP and code generation tasks, incorporating sparse and dense retrieval methods. Prior research in chip design has leveraged fine-tuning open-source LLMs on domain-specific data for improved performance in tasks like Verilog code generation. It also calls for further exploration and enhancement of domain-adapted LLM approaches in chip design.

Electronic Design Automation (EDA) tools have enhanced chip design productivity, yet some time-consuming language-related tasks still need to be completed. LLMs can automate code generation, engineering responses, analysis, and bug triage in chip design. Previous research has explored LLM applications for generating RTL and EDA scripts. Domain-specific LLMs demonstrate superior performance in domain-specific chip design tasks. The aim is to enhance LLM performance while reducing model size. 

The chip design data underwent processing through customised tokenisers, optimising its suitability for analysis. Domain-adaptive continued pretraining procedures were carried out to fine-tune pretrained foundation models, aligning them with the chip design domain. Supervised fine-tuning leveraged domain-specific and general chat instruction datasets to refine model performance. Domain-adapted retrieval models, encompassing both sparse retrieval techniques like TF-IDF and BM25, as well as dense retrieval methods using pretrained models, were harnessed to enhance information retrieval and generation. 

Domain adaptation techniques in ChipNeMo yielded remarkable performance enhancements in LLMs for chip design applications, spanning tasks like engineering chatbots, EDA script generation, and bug analysis. These techniques not only significantly reduced model size but also maintained or improved performance across various design assignments. Domain-adapted retrieval models outshone general-purpose models, showcasing notable improvements—2x better than unsupervised models and a remarkable 30x boost compared to Sentence Transformer models. Rigorous evaluation benchmarks, encompassing multiple-choice queries and code generation assessments, provided quantifiable insights into model accuracy and effectiveness. 

In conclusion, Domain-adapted techniques, such as custom tokenisation, domain-adaptive pretraining, supervised fine-tuning with domain-specific instructions, and domain-adapted retrieval models, marked a substantial enhancement in LLM performance for chip design applications. ChipNeMo models, exemplified by ChipNeMo-13B-Chat, exhibited comparable or superior results to their base models, narrowing the performance gap with more potent LLaMA2 70B models in engineering assistant chatbot, EDA script generation, and bug analysis tasks. 

Check out the Paper. All credit for this research goes to the researchers of this project. Also, don’t forget to join our 32k+ ML SubReddit, 40k+ Facebook Community, Discord Channel, and Email Newsletter, where we share the latest AI research news, cool AI projects, and more.

If you like our work, you will love our newsletter..

We are also on Telegram and WhatsApp.

Large Language Models are the most recent introduction in the Artificial Intelligence community, which has taken the world by storm. These models, due to their incredible capabilities, are being used by everyone, be it researchers, scientists or even students. With their human-imitating potential to answer questions, generate content, summarise text, complete codes and so on, these models have come a long way. 

LLMs are needed in a number of domains, including sentiment analysis, intelligent chatbots, and content creation. These models utilise a lot of computational power, because of which GPU resources are effectively used to increase throughput. This is done by batching several user requests, and to further improve memory efficiency and computing capacity, LLM quantisation techniques are used. However, existing quantisation approaches, like 8-bit weight-activation quantisation, don’t really take advantage of what newer GPUs can accomplish. Since the integer operators on these GPUs are 4-bit, the current quantisation techniques are not designed for maximum efficiency. 

To address this issue, a team of researchers has introduced Atom, a new method that maximises the serving throughput of LLMs. Atom is a low-bit quantisation technique created to increase throughput significantly without sacrificing precision. It uses low-bit operators and low-bit quantisation to reduce memory usage in order to achieve this. It uses a special combination of fine-grained and mixed-precision quantisation to retain excellent accuracy.

The team has shared that Atom has been evaluated in terms of 4-bit weight-activation quantisation configurations when serving. The results demonstrated that Atom can maintain latency within the same goal range while improving end-to-end throughput by up to 7.73 times when compared to the typical 16-bit floating-point (FP16) approach and 2.53 times when compared to 8-bit integer (INT8) quantisation. This makes Atom a viable solution for catering to the increasing demand for their services because it maintains the desired level of response time and greatly increases the speed at which LLMs can process requests.

The researchers have summarised the primary contributions as follows.

LLM serving has been thoroughly analysed as the first step in the study’s performance analysis. The important performance benefits that come from using low-bit weight-activation quantisation approaches have been identified.

A unique and precise low-bit weight-activation quantisation technique called Atom has been presented. 

The team has shared that Atom employs a variety of strategies to guarantee peak performance. It uses mixed precision, which uses reduced precision for the remaining key activations and weights while maintaining accuracy for the former. Fine-grained group quantisation has been used to reduce mistakes during the quantisation process.

Atom employs dynamic activation quantisation, which reduces quantisation mistakes by adjusting to the unique distribution of each input. To further improve overall performance, the method additionally takes care of the KV-cache’s quantisation. 

The research has also proposed an integrated framework for long-term management (LLM) servicing. The team has codesigned an effective inference system, constructing low-bit GPU kernels and showing off Atom’s useful end-to-end throughput and latency in an actual setting.

Atom’s performance has been thoroughly assessed, which shows that Atom greatly increases LLM serving throughput, with throughput gains of up to 7.7x possible at the expense of a minuscule loss of accuracy.

Check out the Paper. All credit for this research goes to the researchers of this project. Also, don’t forget to join our 32k+ ML SubReddit, 40k+ Facebook Community, Discord Channel, and Email Newsletter, where we share the latest AI research news, cool AI projects, and more.

If you like our work, you will love our newsletter..

We are also on Telegram and WhatsApp.

In a recent development, a team of researchers from China have introduced a deep learning model, named circ2CBA, that promises to revolutionize the prediction of binding sites between circular RNAs (circRNAs) and RNA-binding proteins (RBPs). This development holds significant implications for understanding the intricate mechanisms underlying various diseases, particularly cancers.

CircRNAs have garnered substantial attention recently because of their important role in regulating cellular processes and their potential association with various diseases, notably cancer. The interaction between circRNAs and RBPs has emerged as a focal point in this field, as understanding their interplay provides valuable insights into disease mechanisms.

The circ2CBA model, detailed in a recent publication in Frontiers of Computer Science, stands out for its ability to predict binding sites exclusively using sequence information of circRNAs. This marks a big step in making it easier and faster to identify these critical interactions

.

Circ2CBA follows a unique process, which integrates context information between sequence nucleotides of circRNAs and the weight of important positions. The model employs a two-pronged strategy, commencing with the utilization of two layers of Convolutional Neural Networks (CNN) to extract local features from circRNA sequences. This step helps to expand the perception domain, providing a broader scope for analysis.

To understand the fine details between sequence nucleotides, circ2CBA uses a Bidirectional Long Short-Term Memory (BiLSTM) network. It helps the model to recognize complex relationships within the sequence in a better way.

Further augmenting the model’s capabilities is the incorporation of an attention layer, which allocates varying weights to the feature matrix before its input into the two-layer fully connected layer. This meticulous attention to detail ensures that the model can pick up small details in the data.

Ultimately, the prediction outcome is derived by applying a softmax function, resulting in a highly accurate prediction of circRNA-RBP binding sites.

To validate the effectiveness of circ2CBA, the research team sourced circRNA sequences from the CircInteractome database and subsequently selected eight RBPs to construct the dataset. The one-hot encoding method was employed to convert circRNA sequences into a format compatible with the subsequent modelling process.

The results of both comparative and ablation experiments support the efficacy of circ2CBA. Its performance surpasses other existing methods, indicating its potential to advance the field of circRNA-RBP interaction prediction significantly.

Additional motif analysis was conducted to explain the exceptional performance of circ2CBA on specific sub-datasets. The experimental findings provide compelling evidence that circ2CBA represents a powerful and reliable tool for predicting binding sites between circRNAs and RBPs.In conclusion, the circ2CBA deep learning model represents a noteworthy achievement in the study of circRNA-RBP interactions. By using sequence information alone, circ2CBA showcases exceptional accuracy in predicting binding sites, offering new avenues for understanding the role of circRNAs in various diseases, with particular emphasis on cancer. This new method could accelerate progress in the field, driving research towards more precise and efficient interventions in the future.

Check out the Paper and Reference Article. All credit for this research goes to the researchers of this project. Also, don’t forget to join our 32k+ ML SubReddit, 40k+ Facebook Community, Discord Channel, and Email Newsletter, where we share the latest AI research news, cool AI projects, and more.

If you like our work, you will love our newsletter..

We are also on Telegram and WhatsApp.

The computer vision community has been focusing significantly on novel view synthesis (VS) due to its potential to advance artificial reality and enhance a machine’s ability to understand visual and geometric aspects of specific scenarios. State-of-the-art techniques utilizing neural rendering algorithms have achieved photorealistic reconstruction of static scenes. However, current approaches relying on epipolar geometric relationships are better suited for static situations, while real-world scenarios with dynamic elements present challenges to these methods.

Recent works have primarily concentrated on synthesizing views in dynamic settings by using one or more multilayer perceptrons (MLPs) to encode spatiotemporal scene information. One approach involves creating a comprehensive latent representation of the target video down to the frame level. However, the limited memory capacity of MLPs or other representation methods restricts the applicability of this approach to shorter videos despite its ability to deliver visually accurate results.

To address this limitation, researchers from the University of Oxford introduced DynPoint. This unique method doesn’t rely on learning a latent canonical representation to efficiently generate views from longer monocular videos. DynPoint employs an explicit estimation of consistent depth and scene flow for surface points, unlike traditional methods that encode information implicitly. Multiple reference frames’ information is combined into the target frame using these estimates. Subsequently, a hierarchical neural point cloud is constructed from the gathered data, and views of the target frame are synthesized using this hierarchical point cloud.

This aggregation process is supported by learning correspondences between the target and reference frames, aided by depth and scene flow inference. To enable the quick synthesis of the target frame within a monocular video, the researchers provide a representation for aggregating information from reference frames to the target frame. Extensive evaluations of DynPoint’s speed and accuracy in view synthesis are conducted on datasets such as Nerfie, Nvidia, HyperNeRF, iPhone, and Davis. The proposed model demonstrates superior performance in terms of both accuracy and speed, as evidenced by the experimental results.

Check out the Paper. All credit for this research goes to the researchers of this project. Also, don’t forget to join our 32k+ ML SubReddit, 40k+ Facebook Community, Discord Channel, and Email Newsletter, where we share the latest AI research news, cool AI projects, and more.

If you like our work, you will love our newsletter..

We are also on Telegram and WhatsApp.

Table of contents

This Artificial Intelligence tutorial provides basic and intermediate information on concepts of Artificial Intelligence. It is designed to help students and working professionals who are complete beginners. In this tutorial, our focus will be on artificial intelligence, if you wish to learn more about machine learning, you can check out this tutorial for complete beginners tutorial of Machine Learning.

Through the course of this Artificial Intelligence tutorial, we will look at various concepts such as the meaning of artificial intelligence, the levels of AI, why AI is important, it’s various applications, the future of artificial intelligence, and more.

Usually, to work in the field of AI, you need to have a lot of experience. Thus, we will also discuss the various job profiles which are associated with artificial intelligence and will eventually help you to attain relevant experience. You don’t need to be from a specific background before joining the field of AI as it is possible to learn and attain the skills needed. While the terms Data Science, Artificial Intelligence (AI) and Machine learning fall in the same domain and are connected, they have their specific applications and meaning. Simply put, artificial intelligence aims at enabling machines to execute reasoning by replicating human intelligence. Since the main objective of AI processes is to teach machines from experience, feeding the right information and self-correction is crucial.

What is Artificial Intelligence?

The answer to this question would depend on who you ask. A layman, with a fleeting understanding of technology, would link it to robots. If you ask about artificial intelligence to an AI researcher, (s)he would say that it’s a set of algorithms that can produce results without having to be explicitly instructed to do so. Both of these answers are right. So to summarize, Artificial Intelligence is:

An intelligent entity created by humans.

Capable of performing tasks intelligently without being explicitly instructed.

Capable of thinking and acting rationally and humanely.

At the core of Artificial Intelligence, it is a branch of computer science that aims to create or replicate human intelligence in machines. But what makes a machine intelligent? Many AI systems are powered with the help of machine learning and deep learning algorithms. AI is constantly evolving, what was considered to be part of AI in the past may now just be looked at as a computer function. For example, a calculator may have been considered to be a part of AI in the past. Now, it is considered to be a simple function. Similarly, there are various levels of AI, let us understand those.

[embedded content]

Why is Artificial Intelligence Important?

The goal of Artificial Intelligence is to aid human capabilities and help us make advanced decisions with far-reaching consequences. From a technical standpoint, that is the main goal of AI. When we look at the importance of AI from a more philosophical perspective, we can say that it has the potential to help humans live more meaningful lives that are devoid of hard labour. AI can also help manage the complex web of interconnected individuals, companies, states and nations to function in a manner that’s beneficial to all of humanity.

Currently, Artificial Intelligence is shared by all the different tools and techniques have been invented by us over the last thousand years – to simplify human effort, and to help us make better decisions. Artificial Intelligence is one such creation that will help us in further inventing ground-breaking tools and services that would exponentially change how we lead our lives, by hopefully removing strife, inequality and human suffering.

We are still a long way from those kinds of outcomes. But it may come around in the future. Artificial Intelligence is currently being used mostly by companies to improve their process efficiencies, automate resource-heavy tasks, and to make business predictions based on data available to us. As you see, AI is significant to us in several ways. It is creating new opportunities in the world, helping us improve our productivity, and so much more. 

History of Artificial Intelligence

The concept of intelligent beings has been around for a long time and have now found its way into many sectors such as AI in education, automotive, banking and finance, AI healthcare etc. The ancient Greeks had myths about robots as the Chinese and Egyptian engineers built automatons. However, the beginnings of modern AI has been traced back to the time where classical philosophers’ attempted to describe human thinking as a symbolic system. Between the 1940s and 50s, a handful of scientists from various fields discussed the possibility of creating an artificial brain. This led to the rise of the field of AI research – which was founded as an academic discipline in 1956 – at a conference at Dartmouth College, in Hanover, New Hampshire. The word was coined by John McCarthy, who is now considered as the father of Artificial Intelligence.

Despite a well-funded global effort over numerous decades, scientists found it extremely difficult to create intelligence in machines. Between the mid-1970s and 1990s, scientists had to deal with an acute shortage of funding for AI research. These years came to be known as the ‘AI Winters’. However, by the late 1990, American corporations once again were interested in AI. Furthermore, the Japanese government too, came up with plans to develop a fifth-generation computer for the advancement of AI. Finally, In 1997, IBM’s Deep Blue defeated the first computer to beat a world chess champion, Garry Kasparov.

As AI and its technology continued to march – largely due to improvements in computer hardware, corporations and governments too began to successfully use its methods in other narrow domains. The last 15 years, Amazon, Google, Baidu, and many others, have managed to leverage AI technology to a huge commercial advantage. AI, today, is embedded in many of the online services we use. As a result, the technology has managed to not only play a role in every sector, but also drive a large part of the stock market too. 

Today, Artificial Intelligence is divided into sub-domains namely Artificial General Intelligence, Artificial Narrow Intelligence, and Artificial Super Intelligence which we will discuss in detail in this article. We will also discuss the difference between AI and AGI.

Levels of Artificial Intelligence

Artificial Intelligence can be divided into three main levels:

Artificial Narrow Intelligence

Artificial General Intelligence

Artificial Super-intelligence

Artificial Narrow Intelligence (ANI)

Also known as narrow AI or weak AI, Artificial narrow intelligence is goal-oriented and is designed to perform singular tasks. Although these machines are seen to be intelligent, they function under minimal limitations, and thus, are referred to as weak AI. It does not mimic human intelligence; it stimulates human behaviour based on certain parameters. Narrow AI makes use of NLP or natural language processing to perform tasks. This is evident in technologies such as chatbots and speech recognition systems such as Siri. Making use of deep learning allows you to personalise user experience, such as virtual assistants who store your data to make your future experience better. 

Examples of weak or narrow AI:

Siri, Alexa, Cortana

IBMs Watson

Self-driving cars

Facial recognition softwares

Email spam filters 

Prediction tools 

Artificial General Intelligence (AGI)

Also known as strong AI or deep AI, artificial general intelligence refers to the concept through which machines can mimic human intelligence while showcasing the ability to apply their intelligence to solve problems. Scientists have not been able to achieve this level of intelligence yet. Significant research needs to be done before this level of intelligence can be achieved. Scientists would have to find a way through which machines can become conscious through programming a set of cognitive abilities. A few properties of deep AI are-

Recognition

Recall 

Hypothesis testing 

Imagination

Analogy

Implication

It is difficult to predict whether strong AI will continue to advance or not in the foreseeable future, but with speech and facial recognition continuously showing advancements, there is a slight possibility that we can expect growth in this level of AI too. 

Artificial Super-intelligence (ASI)

Currently, super-intelligence is just a hypothetical concept. People assume that it may be possible to develop such an artificial intelligence in the future, but it doesn’t exist in the current world. Super-intelligence can be known as that level wherein the machine surpasses human capabilities and becomes self-aware. This concept has been the muse to several films, and science fiction novels wherein robots who are capable of developing their feelings and emotions can overrun humanity itself. It would be able to build emotions of its own, and hypothetically, be better than humans at art, sports, math, science, and more. The decision-making ability of a super-intelligence would be greater than that of a human being. The concept of artificial super-intelligence is still unknown to us, its consequences can’t be guessed, and its impact cannot be measured just yet. 

Let us now understand the difference between weak AI and strong AI. 

Weak AIStrong AIIt is a narrow application with a limited scope.It is a wider application with a more vast scope.This application is good at specific tasks.This application has an incredible human-level intelligence.It uses supervised and unsupervised learning to process data.It uses clustering and association to process data.Example: Siri, Alexa.Example: Advanced Robotics

Applications of Artificial Intelligence

Artificial intelligence has paved its way into several industries and areas today. From gaming to healthcare, the application of AI has increased immensely. Did you know that the Google Maps applications and facial recognition such as on the iPhone are all using AI technology to function? AI is all around us and is part of our daily lives more than we know it. If you wish to learn more about AI, you can take up the PGP Artificial Intelligence and Machine Learning Course offered by Great Learning. Here are a few applications of Artificial Intelligence.

Best Applications of Artificial Intelligence in 2024

Google’s AI-powered predictions (Google Maps)

Ride-sharing applications (Uber, Lyft)

AI Autopilot in Commercial Flights

Spam filters on Emails

Plagiarism checkers and tools

Facial Recognition

Search recommendations

Voice-to-text features

Smart personal assistants (Siri, Alexa)

Fraud protection and prevention

Now that we know these are the areas where AI is applied. Let us understand these in a more detailed way. Google has partnered with DeepMind to improve the accuracy of traffic predictions. With the help of historical traffic data as well as the live data, they can make accurate predictions through AI technology and machine learning algorithms. An intelligent personal assistant can perform tasks based on commands given by us. It is a software agent and can perform tasks such as sending messages, performing a google search, recording a voice note, chatbots, and more. 

Goals of Artificial Intelligence

So far, you’ve seen what AI means, the different levels of AI, and its applications. But what are the goals of AI? What is the result that we aim to achieve through AI? The overall goal would be to allow machines and computers to learn and function intelligently. Some of the other goals of AI are as follows:

1. Problem-solving: Researchers developed algorithms that were able to imitate the step-by-step process that humans use while solving a puzzle. In the late 1980s and 1990s, research had reached a stage wherein methods had been developed to deal with incomplete or uncertain information. But for difficult problems, there is a need for enormous computational resources and memory power. Thus, the search for efficient problem-solving algorithms is one of the goals of artificial intelligence.

2. Knowledge representation: Machines are expected to solve problems that require extensive knowledge. Thus, knowledge representation is central to AI. Artificial intelligence represents objects, properties, events, cause and effect, and much more. 

3. Planning: One of the goals of AI should be to set intelligent goals and achieve them. Being able to make predictions about how actions will impact change, and what are the choices available. An AI agent will need to assess its environment and accordingly make predictions. This is why planning is important and can be considered as a goal of AI. 

4. Learning: One of the fundamental concepts of AI, machine learning, is the study of computer algorithms that continue to improve over time through experience. There are different types of ML. The commonly known types of are Unsupervised Machine Learning and Supervised Machine Learning. To learn more about these concepts, you can read our blog on what ML means and how it works. 

5. Social Intelligence: Affective computing is essentially the study of systems that can interpret, recognize, and process human efforts. It is a confluence of computer science, psychology, and cognitive science. Social intelligence is another goal of AI as it is important to understand these fields before building algorithms. 

Thus, the overall goal of AI is to create technologies that can incorporate the above goals and create an intelligent machine that can help us work efficiently, make decisions faster, and improve security. 

Jobs in Artificial Intelligence

The demand for AI skills has more than doubled over the last three years, according to Indeed. Job postings in the field of AI have gone up by 119%. The task of training an image-processing algorithm can be done within minutes today, while a few years ago, the task would take hours to complete. When we compare the skilled professionals in the market with the number of job openings available today, we can see a shortage of skilled professionals in the field of artificial intelligence.

Bayesian Networking, Neural nets, computer science (including knowledge about programming languages), physics, robotics, calculus and statistical concepts are a few skills that one must know before deep diving into a career in AI. If you are someone who is looking to build a career in AI, you should be aware of the various job roles available. Let us take a closer look at the different job roles in the world of AI and what skills one must possess for each job role. 

Also Read: Artificial Intelligence Interview Questions 2020

1. Machine Learning Engineer

If you are someone who hails from a background in Data Science or applied research, the role of a Machine Learning Engineer is suitable for you. You must demonstrate an understanding of multiple programming languages such as Python, Java. Having an understanding of predictive models and being able to leverage Natural Language Processing while working with enormous datasets will prove to be beneficial. Being familiar with software development IDE tools such as IntelliJ and Eclipse will help you further advance your career as a machine learning engineer. You will mainly be responsible for building and managing several machine learning projects among other responsibilities.

As an ML engineer, you will receive an annual median salary of $114,856. Companies look for skilled professionals who have a masters degree in the related field and have in-depth knowledge regarding machine learning concepts, Java, Python, and Scala. The requirements will vary depending on the hiring company, but analytical skills and cloud applications are seen as a plus point. 

2. Data Scientist 

As a Data Scientist, your tasks include collecting, analyzing, and interpreting large & complex datasets by leveraging machine learning and predictive analytics tools. Data Scientists are also responsible for developing algorithms that enable collecting and cleaning data for further analysis and interpretation. The annual median salary of a Data Scientist is $120,931, and the skills required are as follows: 

Hive

Hadoop

MapReduce

Pig

Spark

Python

Scala

SQL 

The skills required may vary from company to company, and depending on your experience level. Most hiring companies look for a masters degree or a doctoral degree in the field of data science or computer science. If you’re a Data Scientist who wants to become an AI developer, an advanced computer science degree proves to be beneficial. You must have the ability to understand unstructured data, and have strong analytical and communication skills. These skills are essential as you will work on communicating findings with business leaders. 

3. Business Intelligence Developer 

When you’re looking at the different job roles in AI, it also includes the position of Business Intelligence (BI) developer. The objective of this role is to analyze complex datasets that help us identify business and market trends. A BI developer earns an annual median salary of $92,278. A BI developer is responsible for designing, modelling, and maintaining complex data in cloud-based data platforms. If you are interested to work as a BI developer, you must have strong technical as well as analytical skills.

Having great communication skills is important because you will work on communicating solutions to colleagues who don’t possess technical knowledge. You should also display problem-solving skills. A BI developer is typically required to have a bachelor’s degree in any related field, and work experience will give you additional points too. Certifications are highly desired and are looked at as an additional quality. The skills required for a BI developer would be data mining, SQL queries, SQL server reporting services, BI technologies, and data warehouse design. 

4. Research Scientist 

A research scientist is one of the leading careers in Artificial Intelligence. You should be an expert in multiple disciplines, such as mathematics, deep learning, machine learning, and computational statistics. Candidates must have adequate knowledge concerning computer perception, graphical models, reinforcement learning, and NLP. Similar to Data Scientists, research scientists are expected to have a master’s or doctoral degree in computer science. The annual median salary is said to be $99,809. Most companies are on the lookout for someone who has an in-depth understanding of parallel computing, distributed computing, benchmarking and machine learning. 

5. Big Data Engineer/Architect 

Big Data Engineer/Architects have the best-paying job among all the roles that come under Artificial Intelligence. The annual median salary of a Big Data Engineer/Architect is $151,307. They play a vital role in the development of an ecosystem that enables business systems to communicate with each other and collate data. Compared to Data Scientists, Big data Architects receive tasks related to planning, designing, and developing an efficient big data environment on platforms such as Spark and Hadoop. Companies typically look to hire individuals who demonstrate experience in C++, Java, Python, and Scala. 

Data mining, data visualization, and data migration skills are an added benefit. Another bonus would be a PhD in mathematics or any related computer science field.

Advantages of Artificial Intelligence

Just as it is the case with most things in the world, AI has its pros and cons. First, let us understand the advantages of artificial intelligence and how it has made our lives easier compared to the earlier times. 

Reduction in human error

Available 24×7

Helps in repetitive work

Digital assistance 

Faster decisions

Rational Decision Maker

Medical applications

Improves Security

Efficient Communication

Let’s take a closer look at each of the aforementioned points. 

1. Reduction in human error

All decisions taken in an AI model are taken from previously gathered information after having applied a set of algorithms. This enables the errors to be reduced, and the chances of accuracy increase with a greater degree of accuracy. In the case of humans performing any task, there’s always a slight chance of error. Since we are capable of making errors, it is better to make use of programs and algorithms through AI as they lower the chance of errors. 

2. Available 24×7

Artificial intelligence models are built to work 24/7 without any breaks or boredom. When compared to an average human who can work for six to eight hours in a day, this is significantly more efficient. Human beings do not have the capacity to work for longer durations as we would require rest and time to rejuvenate. Thus, AI is available 24/7 and improves efficiency to a greater extent. 

3. Helps in repetitive work

Artificial Intelligence can productively automate mundane human tasks. It can help us in becoming increasingly creative – right from sending a thank you mail to decluttering or answering queries. It can also help us in verifying documents. A repetitive task such as making food in a restaurant or a factory can be ruined because humans become tired or uninterested after a long duration of work. AI can help us in performing these repetitive tasks efficiently and without error. 

4. Digital assistance

Several organisations who are highly advanced make use of digital assistants to interact with users. Doing so helps the organisation to save costs on human resources. Digital assistants such as Chatbots are typically used in an organisations website to answer user queries. It also provides a smooth functioning interface and good user experience. Chatbots are a great example of the same. Read here to know more about how to build an AI Chatbot.

5. Faster decisions 

AI, alongside other such technologies, can help machines take faster decisions when compared to an average human being. This helps in carrying out actions quickly. This is because, while making a decision, humans tend to analyse factors through emotions as opposed to AI-powered machines that deliver programmed results quickly.

6. Rational Decision Maker

We as humans may have evolved to a great extent technologically, but when it comes to decision making, we still allow our emotions to take over. In certain situations, it is really important to take quick, efficient and logical decisions without our emotions coming into the picture. AI-powered decision making is controlled by AI algorithms, and thus, there is no scope for any emotional discrepancy. Rational decisions with the help of AI ensures that efficiency will not be affected, and also increases an organisation’s productivity level. 

7. Medical applications

Among all other advantages of AI, one of the greatest applications in its use in the medical field. Doctors can assess their patients’ health risks with the help of AI-powered medical applications. Radiosurgery is being used to operate on tumors in such a way that it won’t damage surrounding tissues and cause any additional issues. Medical professionals have been trained to use AI for surgery. They can also help in efficiently detecting and monitoring various neurological disorders and stimulate the brain functions. 

8. Improves Security

As technology continues to advance, there is a higher chance of people using it for unethical reasons such as fraud or identity theft. If used in the right manner and for the right reasons, AI can prove to be a great resource in improving our organisation’s security. AI can be used to protect our data and finances. AI is being implemented majorly in the field of cybersecurity. It has transformed our ability to secure our personal data against any cyber-threats or attacks of any form. Read further to know about AI in Cybersecurity and how it helps, here.

9. Efficient Communication 

People from different parts of the world speak different languages and thus, find it hard to communicate with each other. When we look at the past, we see how human translators would help people communicate with each other if the other person did not understand the same language as us. Such problems do not occur if we make use of AI. Natural Language Processing allows systems to translate words from one natural language to another, thus eliminating the middleman. One of the best examples of this is Google translate, and how it has advanced over time. Now, it provides audio examples of how words/sentences should be pronounced. Thus, improving our accuracy and ability to communicate effectively.

Disadvantages of Artificial Intelligence

Now that we have understood the advantages of AI, let us take a look at a few disadvantages. 

Cost overruns

Dearth of talent

Lack of practical products

Lack of standards in software development

Potential for misuse

Highly dependent on machines

Requires Supervision

Let’s take a closer look at the disadvantages of AI. 

1. Cost overruns

The scale of operations of an AI-powered model when compared to software development is massively higher. Due to this, the resources required increase at a much higher rate. This pushes the cost of operations to a higher level.

2. Dearth of talent 

AI is still a field which is developing. Thus, finding professionals who are equipped with all the required skills is not easy. There is a gap between the number of jobs available in the field of AI vs the skilled workforce in the field. Hiring someone who possesses all the necessary skills further increases the costs incurred by an organisation.

3. Lack of standards in software development

The true value of Artificial Intelligence lays in collaboration when different AI systems come together to form a bigger, more valuable application. But a lack of standards in AI software development means that it’s difficult for different systems to ‘talk’ to each other. Artificial Intelligence software development itself is slow and expensive because of this, which further acts as an impediment to AI development.

4. Potential for Misuse

AI has the potential to achieve great things, and has massive power in the market today. Unfortunately, with great power comes the potential of misuse. If the power of AI falls in the hands of a person who has unethical motives, there is a higher chance of misuse.

5. Highly dependent on machines

Applications such as Siri and Alexa have become part of our everyday lives. We are highly dependent on these applications and receive assistance from these applications, thus reducing our creative ability. We are becoming highly dependent on machines and losing our on learning simple skills, thus becoming lazier. 

6. Requires Supervision

Making use of AI algorithms has a lot of advantages and is highly efficient. But it also requires constant assistance and supervision. These algorithms cannot work without us programming them and checking if they are functioning in the right manner or not. One example is Microsoft’s AI chat-bot named ‘Tay’. Tay was modelled to speak like a teenage girl by learning through online conversations. But since it was programmed to learn basic conversational skills and didn’t know the difference between right and wrong, it went ahead and tweeted highly political and incorrect information because of internet trolls.

Future of Artificial Intelligence

We have always been fascinated by technological changes. Currently, we are living amidst the greatest AI advancements in our history. Artificial Intelligence has emerged to be the net greatest advancement in the field of technology. This has not only impacted the future of every industry, but has also acted as the driver of emerging technologies such as big data, robotics and IoT. At that rate at which AI is advancing, there is no doubt that it will continue to flourish in the future. Thus, we can say that AI is a great field to enter as of 2020. With the advancement of AI and its technologies, there will be a greater need for skilled professionals in this area.

An AI certification will give you an edge over other participants in the industry. As Facial Recognition, AI in Healthcare, Chat-bots continue to show growth, now would be the right time to work towards building a successful AI career. Virtual assistants are already part of our everyday life without us knowing it. Self-driving cars by Tech giants like Tesla have shown us a glimpse of what the future will look like. There are so many more advancements to be discovered, this is only the beginning. According to the World Economic Forum, 133 million new Artificial Intelligence jobs are said to be created by Artificial Intelligence by the year 2022. The future of AI is definitely bright.

A simple Artificial Intelligence mini-project

Before moving on to the project, I would suggest going through this Machine learning Tutorial if you are not familiar with Machine learning at all. It would also help you with this project if you know about the Logistic Regression algorithm.

Zoo Animal Classification

In this mini-project, we will use different algorithms that come under the Machine learning domain of Artificial Intelligence to classify animals in a zoo, based on their attributes. We are going to use this dataset from Kaggle which consists of 101 animals from a zoo. There are 16 variables with various traits to describe the animals. The 7 Class Types are: Mammal, Bird, Reptile, Fish, Amphibian, Bug and Invertebrate.

The purpose of this dataset is to be able to predict the classification of the animals based on the variables. You can also find the information about the various attributes used in this dataset from the download page linked here.

import numpy as np
import pandas as pd
from sklearn.model_selection import train_test_split
df = pd.read_csv(r’/content/zoo.csv’)
df.head()

Output:

features.remove(‘class_type’)
features.remove(‘animal_name’)
X = df[features].values.astype(np.float32)
Y = df.class_type
X_train, X_test, Y_train, Y_test = train_test_split(X, Y, test_size = 0.5, random_state = 0)

from sklearn.linear_model import LogisticRegression
model = LogisticRegression()
model.fit(X_train, Y_train)
print(“training accuracy :”, model.score(X_train, Y_train))
print(“testing accuracy :”, model.score(X_test, Y_test))

Output:
training accuracy : 1.0
testing accuracy : 0.9215686274509803

As you can see, the model performed exceptionally well by getting 92% accuracy on the testing data. Now, if you are given the attributes of any of the animal in the above dataset, you can classify it with the help of the above model.

Will AI reduce jobs in future?

AI is still developing. There is a huge scope for improvement and advancements in the field of AI, and although there might be some amount of upskilling required to keep up with the changing trends, AI will most likely not replace or reduce jobs in the future. In fact, a study by Gartner suggests that AI-related jobs will reach two million net-new jobs by the year 2025. The adoption of AI will help in making tasks easier for an organization. To stay relevant in the constantly changing world, it is necessary to upskill and learn these new concepts.

How AI Works?

Building an AI system is a careful process of reverse-engineering human traits and capabilities in a machine, and using it’s computational prowess to surpass what we are capable of. Artificial Intelligence can be built over a diverse set of components and will function as an amalgamation of:

Philosophy

Mathematics

Economics

Neuroscience

Psychology

Computer Engineering

Control Theory and Cybernetics

Linguistics

How is artificial intelligence used in robotics?

Artificial Intelligence and Robotics are usually seen as two different things. AI involves programming intelligence whereas robotics involves building physical robots. However, the two concepts are correlated. Robotics does use AI techniques and algorithms and AI bridges the gap between the two. These robots can be controlled by AI programs.

Why is artificial intelligence important?

From music recommendations, map directions, mobile banking to fraud prevention, AI and other technologies have taken over. AI is important because of a number of reasons. There are several advantages to AI, such as, Reduction in human error, available 24×7, helps in repetitive work, digital assistance, faster decisions and more.

What are weak methods in AI?

Weak AI is a narrow application with a limited scope. It uses supervised and unsupervised learning to process data. Example: Siri, Alexa.

What are the branches of AI?

Artificial Intelligence can be divided mainly into six branches. They are, Machine learning, neural networks, Deep Learning, Computer Vision, Natural language processing, Cognitive Computing. 

How can I start learning Artificial Intelligence?

To learn Artificial Intelligence, you need to have skills such as Math, Science, and Computer Science. You can also opt for some online tutorials and learn Artificial Intelligence from the comfort of your home.

What are the 4 types of AI? 

 The four typical types of Artificial Intelligence are Reactive Machines, Limited Memory, Theory of Mind, and Self Aware.

What are the basic things to learn Artificial Intelligence?

The basics of Artificial Intelligence are Advanced Math and Stats, programming language, Machine Learning, and a lot of patience. You must know that Artificial Intelligence and Machine Learning include machine learning, python code, computer science, natural language processing, data science, math, psychology, neuroscience, and many other disciplines.

Is AI difficult to learn?

 Artificial Intelligence is not tough; however, you would be required to spend time on it. The more the number of projects you work on, the better you will get at it. Along with skills, you need the determination to learn AI.

This brings us to the end of the Artificial Intelligence tutorial. Here is a free course about AIML that can help you to make your foundations much stronger.