Рубрика: AI

Image by Editor

The onset of Graphical Processing Units (GPUs), and the exponential computing power they unlock, has been a watershed moment for startups and enterprise businesses alike.

GPUs provide impressive computational power to perform complex tasks that involve technology such as AI, machine learning, and 3D rendering.

However, when it comes to harnessing this abundance of computational power, the tech world stands at a crossroads in terms of the ideal solution. Should you build a dedicated GPU machine or utilize the GPU cloud?

This article delves into the heart of this debate, dissecting the cost implications, performance metrics, and scalability factors of each option.

What is a GPU?

GPUs (Graphical Processing Units) are computer chips that are designed to rapidly render graphics and images by completing mathematical calculations almost instantaneously. Historically, GPUs were often associated with personal gaming computers, but they are also used in professional computing, with advancements in technology requiring additional computing power.

GPUs were initially developed to reduce the workload being placed on the CPU by modern, graphic-intensive applications, rendering 2D and 3D graphics using parallel processing, a method that involves multiple processors handling different parts of a single task.

In business, this methodology is effective in accelerating workloads and providing enough processing power to enable projects such as artificial intelligence (AI) and machine learning (ML) modeling.

GPU Use Cases

GPUs have evolved in recent years, becoming much more programmable than their earlier counterparts, allowing them to be used in a wide range of use cases, such as:

• Rapid rendering of real-time 2D and 3D graphical applications, using software like Blender and ZBrush
• Video editing and video content creation, especially pieces that are in 4k, 8k or have a high frame rate
• Providing the graphical power to display video games on modern displays, including 4k.
• Accelerating machine learning models, from basic image conversion to jpg to deploying custom-tweaked models with full-fledged front-ends in a matter of minutes
• Sharing CPU workloads to deliver higher performance in a range of applications
• Providing the computational resources to train deep neural networks
• Mining cryptocurrencies such as Bitcoin and Ethereum

Focusing on the development of neural networks, each network consists of nodes that each perform calculations as part of a wider analytical model.

GPUs can enhance the performance of these models across a deep learning network thanks to the greater parallel processing, creating models that have higher fault tolerance. As a result, there are now numerous GPUs on the market that have been built specifically for deep learning projects, such as the recently announced H200.

Building a GPU Machine

Many businesses, especially startups choose to build their own GPU machines due to their cost-effectiveness, while still offering the same performance as a GPU cloud solution. However, this is not to say that such a project does not come with challenges.

In this section, we will discuss the pros and cons of building a GPU machine, including the expected costs and the management of the machine which may impact factors such as security and scalability.

Why Build Your Own GPU Machine?

The key benefit of building an on-premise GPU machine is the cost but such a project is not always possible without significant in-house expertise. Ongoing maintenance and future modifications are also considerations that may make such a solution unviable. But, if such a build is within your team’s capabilities, or if you have found a third-party vendor that can deliver the project for you, the financial savings can be significant.

Building a scalable GPU machine for deep learning projects is advised, especially when considering the rental costs of cloud GPU services such as Amazon Web Services EC2, Google Cloud, or Microsoft Azure. Although a managed service may be ideal for organizations looking to start their project as soon as possible.

Let’s consider the two main benefits of an on-premises, self-build GPU machine, cost and performance.

Costs

If an organization is developing a deep neural network with large datasets for artificial intelligence and machine learning projects, then operating costs can sometimes skyrocket. This can hinder developers from delivering the intended outcomes during model training and limit the scalability of the project. As a result, the financial implications can result in a scaled-back product, or even a model that is not fit for purpose.

Building a GPU machine that is on-site and self-managed can help to reduce costs considerably, providing developers and data engineers with the resources they need for extensive iteration, testing, and experimentation.

However, this is only scratching the surface when it comes to locally built and run GPU machines, especially for open-source LLMs, which are growing more popular. With the advent of actual UIs, you might soon see your friendly neighborhood dentist run a couple of 4090s in the backroom for things such as insurance verification, scheduling, data cross-referencing, and much more.

Performance

Extensive deep learning and machine learning training models/ algorithms require a lot of resources, meaning they need extremely high-performing processing capabilities. The same can be said for organizations that need to render high-quality videos, with employees requiring multiple GPU-based systems or a state-of-the-art GPU server.

Self-built GPU-powered systems are recommended for production-scale data models and their training, with some GPUs able to provide double-precision, a feature that represents numbers using 64 bits, providing a larger range of values and better decimal precision. However, this functionality is only required for models that rely on very high precision. A recommended option for a double-precision system is Nvidia’s on-premise Titan-based GPU server.

Operations

Many organizations lack the expertise and capabilities to manage on-premise GPU machines and servers. This is because an in-house IT team would need experts who are capable of configuring GPU-based infrastructure to achieve the highest level of performance.

Furthermore, his lack of expertise could lead to a lack of security, resulting in vulnerabilities that could be targeted by cybercriminals. The need to scale the system in the future may also present a challenge.

Using the GPU Cloud

On-premises GPU machines provide clear advantages in terms of performance and cost-effectiveness, but only if organizations have the required in-house experts. This is why many organizations choose to use GPU cloud services, such as Saturn Cloud which is fully managed for added simplicity and peace of mind.

Cloud GPU solutions make deep learning projects more accessible to a wider range of organizations and industries, with many systems able to match the performance levels of self-built GPU machines. The emergence of GPU cloud solutions is one of the main reasons people are investing in AI development more and more, especially open-source models like Mistral, whose open-source nature is tailor-made for ‘rentable vRAM’ and running LLMs without depending on larger providers, such as OpenAI or Anthropic.

Costs

Depending on the needs of the organization or the model that is being trained, a cloud GPU solution could work out cheaper, providing the hours it is needed each week are reasonable. For smaller, less data-heavy projects, there is probably no need to invest in a costly pair of H100s, with GPU cloud solutions available on a contractual basis, as well as in the form of various monthly plans, catering to the enthusiast all the way to enterprise.

Performance

There is an array of CPU cloud options that can match the performance levels of a DIY GPU machine, providing optimally balanced processors, accurate memory, a high-performance disk, and eight GPUs per instance to handle individual workloads. Of course, these solutions may come at a cost but organizations can arrange hourly billing to ensure they only pay for what they use.

Operations

The key advantage of a cloud GPU over a GPU build is in its operations, with a team of expert engineers available to assist with any issues and provide technical support. An on-premise GPU machine or server needs to be managed in-house or a third-party company will need to manage it remotely, coming at an additional cost.

With a GPU cloud service, any issues such as a network breakdown, software updates, power outages, equipment failure, or insufficient disk space can be fixed quickly. In fact, with a fully managed solution, these issues are unlikely to occur at all as the GPU server will be optimally configured to avoid any overloads and system failures. This means IT teams can focus on the core needs of the business.

Conclusion

Choosing between building a GPU machine or using the GPU cloud depends on the use case, with large data-intensive projects requiring additional performance without incurring significant costs. In this scenario, a self-built system may offer the required amount of performance without high monthly costs.

Alternatively, for organizations who lack in-house expertise or may not require top-end performance, a managed cloud GPU solution may be preferable, with the machine’s management and maintenance taken care of by the provider.

Nahla Davies is a software developer and tech writer. Before devoting her work full time to technical writing, she managed—among other intriguing things—to serve as a lead programmer at an Inc. 5,000 experiential branding organization whose clients include Samsung, Time Warner, Netflix, and Sony.

More On This Topic

• Using RAPIDS cuDF to Leverage GPU in Feature Engineering
• 11 Best Practices of Cloud and Data Migration to AWS Cloud
• Building Machine Learning Pipelines using Snowflake and Dask
• Super Charge Python with Pandas on GPUs Using Saturn Cloud
• Create and Deploy Dashboards using Voila and Saturn Cloud
• eBook: A Practical Guide to Using Third-Party Data in the Cloud

Image by Author

For those diving into the world of computer science or needing a touch-up on their probability knowledge, you’re in for a treat. Stanford University has recently updated its YouTube playlist on its CS109 course with new content!

The playlist comprises 29 lectures to provide you with gold-standard knowledge of the basics of probability theory, essential concepts in probability theory, mathematical tools for analyzing probabilities, and then ending data analysis and Machine Learning.

So let’s get straight into it…

Lecture 1: Counting

Link: Counting

Learn about the history of probability and how it has helped us achieve modern AI, with real-life examples of developing AI systems. Understand the core counting phases, counting with ‘steps’ and counting with ‘or’. This includes areas such as artificial neural networks and how researchers would use probability to build machines.

Lecture 2: Combinatorics

Link: Combinatorics

The second lecture goes into the next level of seriousness counting — this is called Combinatorics. Combinatorics is the mathematics of counting and arranging. Dive into counting tasks on n objects, through sorting objects (permutations), choosing k objects (combinations), and putting objects in r buckets.

Lecture 3: What is Probability?

Link: What is Probability?

This is where the course really starts to dive into Probability. Learn about the core rules of probability with a wide range of examples and a touch on the Python programming language and its use with probability.

Lecture 4: Probability and Bayes

Link: Probability and Bayes

In this lecture, you will dive into learning how to use conditional probabilities, chain rule, the law of total probability and Bayes theorem.

Lecture 5: Independence

Link: Independence

In this lecture, you will learn about probability in respect of it being mutually exclusive and independent, using AND/OR. The lecture will go through a variety of examples for you to get a good grasp.

Lecture 6: Random Variables and Expectations

Link: Random Variables and Expectations

Based on the previous lectures and your knowledge of conditional probabilities and independence, this lecture will dive into random variables, use and produce the probability mass function of a random variable, and be able to calculate expectations.

Lecture 7: Variance Bernoulli Binomial

Link: Variance Bernoulli Binomial

You will now use your knowledge to solve harder and harder problems. Your goal for this lecture will be to recognise and use Binomial Random Variables, Bernoulli Random Variables, and be able to calculate the variance for random variables.

Lecture 8: Poisson

Link: Poisson

Poisson is great when you have a rate and you care about the number of occurrences. You will learn about how it can be used in different aspects along with Python code examples.

Lecture 9: Continuous Random Variables

Link: Continuous Random Variables

The goals of this lecture will include being comfortable using new discrete random variables, integrating a density function to get a probability, and using a cumulative function to get a probability.

Lecture 10: Normal Distribution

Link: Normal Distribution

You may have heard this about normal distribution before, in this lecture, you will go through a brief history of normal distribution, what it is, why it is important and practical examples.

Lecture 11: Joint Distributions

Link: Joint Distributions

In the previous lectures, you will have worked with 2 random variables at most, the next step of learning will be to go into any given number of random variables.

Lecture 12: Inference

Link: Inference

The learning goal of this lecture is how to use multinomials, appreciate the utility of log probabilities, and be able to use the Bayes theorem with random variables.

Lecture 13: Inference II

Link: Inference II

The learning goal continues from the last lecture of combining Bayes theorem with random variables.

Lecture 14: Modeling

Link: Modelling

In this lecture, you will take everything you have learned so far and put it into perspective about real-life problems — probabilistic modelling. This is taking a whole bunch of random variables being random together.

Lecture 15: General Inference

Link: General Inference

You will dive into general inference, and in particular, learn about an algorithm called rejection sampling.

Lecture 16: Beta

Link: Beta

This lecture will go into the random variables of probabilities which are used to solve real-world problems. Beta is a distribution for probabilities, where its range values between 0 and 1.

Lecture 17: Adding Random Variables

Link: Adding Random Variables I

At this point of the course, you will be learning about deep theory and adding random variables is an introduction to how to attain results of the theory of probability.

Lecture 18: Central Limit Theorem

Link: Central Limit Theorem

In this lecture, you will dive into the central limit theorem which is an important element in probability. You will go through practical examples so that you can grasp the concept.

Lecture 19: Bootstrapping and P-Values

Link: Bootstrapping and P-Values I

You will now move into uncertainty theory, sampling and bootstrapping which is inspired by the central limit theorem. You will go through practical examples.

Lecture 20: Algorithmic Analysis

Link: Algorithmic Analysis

In this lecture, you will dive a bit more into computer science with an in-depth understanding of the analysis of algorithms, which is the process of finding the computational complexity of algorithms.

Lecture 21: M.L.E.

Link: M.L.E.

This lecture will dive into parameter estimation, which will provide you with more knowledge on machine learning. This is where you take your knowledge of probability and apply it to machine learning and artificial intelligence.

Lecture 22: M.A.P.

Link: M.A.P.

We’re still at the stage of taking core principles of probability and how it applied to machine learning. In this lecture, you will focus on parameters in machine learning regarding probability and random variables.

Lecture 23: Naive Bayes

Link: Naive Bayes

Naive Bayes is the first machine learning algorithm you will learn about in depth. You will have learnt about the theory of parameter estimation, and now will move on to how core algorithms such as Naive Bayes lead to ideas such as neural networks.

Lecture 24: Logistic Regression

Link: Logistic Regression

In this lecture, you will dive into a second algorithm called Logistic regression which is used for classification tasks, which you will also learn more about.

Lecture 25: Deep Learning

Link: Deep Learning

As you’ve started to dive into machine learning, this lecture will go into further detail about deep learning based on what you have already learned.

Lecture 26: Fairness

Link: Fairness

We live in a world where machine learning is being implemented in our day-to-day lives. In this lecture, you will look into the fairness around machine learning, with a focus on ethics.

Lecture 27: Advanced Probability

Link: Advanced Probability

You have learnt a lot about the basics of probability and have applied it in different scenarios and how it relates to machine learning algorithms. The next step is to get a bit more advanced about probability.

Lecture 28: Future of Probability

Link: Future of Probability

The learning goal for this lecture is to learn about the use of probability and the variety of problems that probability can be applied to solve these problems.

Lecture 29: Final Review

Link: Final Review

And last but not least, the last lecture. You will go through all the other 28 lectures and touch on any uncertainties.

Wrapping it up

Being able to find good material for your learning journey can be difficult. This probability for computer science course material is amazing and can help you grasp concepts of probability that you were unsure of or needed a touch up.

Nisha Arya is a Data Scientist and Freelance Technical Writer. She is particularly interested in providing Data Science career advice or tutorials and theory based knowledge around Data Science. She also wishes to explore the different ways Artificial Intelligence is/can benefit the longevity of human life. A keen learner, seeking to broaden her tech knowledge and writing skills, whilst helping guide others.

More On This Topic

• Start a career in Computer Science with Penn’s Master in Computer…
• How our Obsession with Algorithms Broke Computer Vision: And how…
• KDnuggets News, July 6: 12 Essential Data Science VSCode…
• Free From Stanford: Machine Learning with Graphs
• Master Transformers with This Free Stanford Course!
• The Importance of Probability in Data Science

When OpenAI board members ousted Sam Altman, his investors, employees, customers, and technology partners stood with him, demanding his return to the company.

In the Indian context, we went scouting for star tech and startup leaders who too would enjoy a solid backing from the entire ecosystem, if they were to (hypothetically) encounter Altman’s unfortunate ordeal.

Sridhar Vembu

Early this year, when startups like Dukaan, and BYJU’s were on a mass firing spree because of overhiring during a pandemic, Sridhar Vembu, co-founder and chief executive officer at Zoho, announced that they would not be laying off employees. Instead, the company would be expanding its geographical presence with operations in Coimbatore, Madurai and more.

The company is committed to investing in R&D and people, having opened two hub offices and planning additional expansions in Tamil Nadu and Uttar Pradesh.

Over time, Vembu has built a strong culture of R&D during this period with a focus on reinvention and has gathered the admiration of employees, investors and the common people. His distinctive leadership style, emphasising autonomy, intrapreneurship, and a rejection of traditional corporate norms, has shaped this ecosystem leading to the founding of several successful startups by former Zoho staff.

This culture has incubated a diverse range of startups, such as SuperOps.ai, led by Arvind Parthiban, who lauds Vembu’s pride in Indian roots and resistance to the notion of the West’s superiority. Prabhu Ramachandran, a long-time Zoho employee, co-founded Facilio, showcasing how Zoho’s principles of extreme ownership inspires entrepreneurial journeys.

SurveySparrow, Bevywise, and Brand Maxima are other ventures reflecting Zoho’s ethos of learning by doing and focus on customer-centric solutions.

Vijay Shekhar Sharma

Paytm founder Vijay Shekhar Sharma is known for his all-inclusive approach to employees. He leads Paytm with a focus on growth, efficiency, and innovation. His approach emphasises quick decision-making, fostering creativity, and resilience in facing challenges. This leadership style has been pivotal in shaping Paytm’s success and inspiring his team.

Back in 2018, when it had just been three years since the fintech became a unicorn, over 100 past and current employees became millionaires following the sale of company shares worth Rs 500 crore. Interestingly, his office clerk earned Rs 20 lakh from the stock sale, demonstrating a leadership style that values and rewards all levels of staff, not just executives.

His path to founding Paytm, a leading digital payment platform in India, is marked by his early start in the 2000s with One97 Communications. Sharma’s foresight in recognising the potential of digital payments in India was particularly evident during the 2016 demonetisation. He was backed by investing behemoths like Alibaba and SoftBank.

Under his guidance, Paytm broadened its scope to include various financial services, thriving due to Sharma’s emphasis on innovation, customer-centric strategies, and an in-depth understanding of the Indian market.

For over 20 years in the industry, he has been revered by colleagues, peers, and others for his humility, lack of pretence, and kind approach.

Amit Lakhotia

Born out of the Paytm camp is Amit Lakhotia, who served as the VP of business in the company from 2012 to 2015. He then moved to Tokopedia when, during his time, the company saw its valuation rise from $500-$600 million to $7 billion. Eventually, he left to become co-founder and chief executive at auto tech platform Park+ to solve parking challenges in 2019.

In four years since its inception, Park+ has successfully raised $54.3 million through six funding rounds, including a significant Series C round on January 10, 2023, with support from 21 investors like Matrix Partners India, Kunal Khattar, and more.

A pivotal moment was the Series B round in November 2021, securing $25 million led by Sequoia Capital India and showcasing investor confidence. The Series C round garnered around $18 million, boosting Park+’s valuation to $340 million from $155 million, reflecting its fast growth and potential in the market.

The company also expanded by acquiring cartech startup otopark in June 2022.

Kunal Shah

Alongside Vembu, Sharma and Lakhotia, we also have Kunal Shah, founder and chief of CRED. Shah’s story started with him co-founding FreeCharge in 2010 and then selling it to Snapdeal in 2015. Shah founded CRED in 2018, which became a unicorn in 2021.

He has a distinct leadership style that is built on a relaxed open culture, promoting creativity and communication among employees, even engaging in fun meme-making activities of leadership. Influenced by his philosophical background, Shah embraces uncertainty in entrepreneurship, emphasising truth-seeking and continuous improvement.

Bhavish Aggarwal

Ola Cabs founder Bhavish Aggarwal’s venture into entrepreneurship was sparked by a negative car rental experience, highlighting the deficiencies in India’s transport sector.

Launching Ola Cabs in 2010, Aggarwal navigated various hurdles, such as investor scepticism and enlisting drivers, using his determination to secure essential funding for growth. He worked with Microsoft Research before founding Ola.

Under his guidance, Ola expanded rapidly, competing against giants like Uber and branching into new areas, including food delivery and electric vehicles. Now, he is venturing into implementing AI.

Aggarwal’s leadership style is defined by his visionary and adaptable approach, a commitment to learning from mistakes, and a focus on making impactful, and technology-driven advancements.

Nithin Kamath

We can’t talk about inspiring leaders and not mention Nithin Kamath, the Zerodha boss, who values being respected over being liked or feared as a leader. Zerodha’s business model, under Kamath’s leadership, is built from the ground up without external funding, leveraging technology and a unique ‘brokerage-free’ model.

The firm adopts a customer-first philosophy, treating customers fairly and avoiding intrusive marketing. Building long-term trust and credibility with clients is a key focus, rather than short-term aggressive marketing strategies. He also guides budding entrepreneurs emphasising passion-driven problem-solving and gaining relevant experience.

Read more: Can Mira Murati Save OpenAI?

The post If Sam Altman was from India… appeared first on Analytics India Magazine.

Pfizer, a major global player in the pharmaceutical and biotech industry since its establishment in 1849 in New York, has been collaborating with AWS to harness the vast capabilities of AI. In the past year and a half, Pfizer introduced approximately 19 medical products, including vaccines, with digital data and AI contributing to the rapid market entry of 13 of these. A pivotal element in Pfizer’s achievements is its alliance with AWS, initiated in 2019 with the development of a dedicated scientific data cloud (SDC).

“AWS cloud allows scientists to rapidly access historical molecule and compound data, a process that previously took weeks or months,” said Lidia Fonseca, chief digital and technology officer at Pfizer, at AWS re: Invent in Las Vegas.

This enhancement in data accessibility and analysis has accelerated the company’s research and development, particularly in computational research and AI algorithms that assist in identifying and designing promising new molecules.

The partnership took a significant leap forward in 2021 when Pfizer undertook a massive digital transformation. The company shifted a substantial portion of its coding and server infrastructure to the AWS cloud. This transition involved moving 1000 applications and 8000 servers in just 42 weeks, marking one of the fastest migrations for a company of Pfizer’s size.

This shift resulted in an annual savings of $37 million and significantly reduced the health tech giant’s carbon footprint, eliminating approximately 4700 tons of CO2 emissions annually. This transition to the AWS cloud was a cost-saving measure and a strategic move that empowered the company to innovate with greater speed and scale.

Fighting COVID-19 with AWS

“The value of our partnership became particularly evident during the COVID-19 pandemic as AWS’s support was instrumental in various aspects of the vaccine’s development, from manufacturing to clinical trials,” added Fonseca.

When they needed to rapidly scale up computing capacity for intensive data analysis or expedite the submission of crucial data to regulatory bodies, AWS provided the necessary resources. This collaboration was pivotal in enabling the latter to develop and obtain FDA emergency use authorisation for its COVID-19 vaccine in a record 269 days, typically taking 8 to 10 years.

AWS swiftly increased high-performance computing resources for Pfizer, offering a significant expansion in cloud computing capacity, enabling them to perform detailed analyses crucial for vaccine development. Additionally, when fast data submission for FDA approval was required, AWS provided immediate extra computing power, facilitating Pfizer’s timely scientific progress.

Pfizer and AWS’s ongoing collaboration, utilising AI for proactive supply chain management, was vital during events like Hurricane Ian, maintaining essential medicine and vaccine distribution. Before the pandemic, they produced 220 million vaccine doses annually, which increased to 4 billion COVID-19 vaccine Comirnaty doses in 2022, a significant jump attributed to this technological partnership.

Leveraging Generative AI for Cancer Treatment

Looking ahead, Pfizer is progressively leveraging generative AI in various domains, with a particular focus on using AWS’s Bedrock and Sagemaker. This collaboration is expected to yield substantial annual cost savings, estimated between $760 million and $4 billion. Specifically, the company is exploring 17 use cases, with some priority areas potentially saving $750 million to $1 billion each year.

The company has integrated its internal generative AI platform, VOX, with AWS’s cloud services and machine learning models for them to prototype and explore use cases rapidly in research and development, manufacturing, marketing, and more.

Notably, VOX allows for secure innovation with access to large language models, including Amazon Bedrock and Amazon SageMaker. It is being used to create first drafts of patent applications and generate medical and scientific content, significantly speeding up the process and enabling Pfizer to deliver breakthroughs to patients faster.

However, their use of AI extends beyond cost savings and operational efficiency. In the realm of oncology, for example, AI is being used to streamline and accelerate the identification of new treatment targets. This process, which is mainly manual and time-consuming, is being transformed by AI’s ability to collate and analyse relevant data and scientific content rapidly.

Fonseca added that the use of generative AI in manufacturing is also yielding significant improvements, enhancing process optimisation and real-time anomaly detection, which are crucial for maintaining quality and efficiency in pharmaceutical production.

Pfizer is also set to acquire leading biotech firm Seagen, which specialises in innovative cancer treatments. “We aspire to replicate with cancer treatment the success we achieved with our COVID response. Our collaboration with AWS plays a critical role in maintaining this momentum, driving our goal to positively impact a million lives every year, and not only in times of a pandemic,” Fonseca concluded.

AWS’s Big Bets on Healthcare

Besides Pfizer, AWS is also collaborating with other prominent healthcare and pharmaceutical companies like Amgen, Moderna, BioNTech, Gilead, and Johnson & Johnson to enhance drug discovery, manufacturing, and AI integration in healthcare.

AWS’s role extends to data management, partnering with Dedalus, Change Healthcare, and Visage Imaging for data migration and with CrowdStrike and TrendMicro for data protection. They also work with GE Healthcare and Infor on data unification and with DataBricks and Philips in innovating AI and machine learning applications in healthcare.

Additional partnerships with 3M, Hyland Software, NextGen, and Pariveda focus on improving EHR, virtual care, interoperability, and population health, showcasing AWS’s dedication to advancing healthcare technology.

Read more: [Exclusive] AWS’ Generative AI Play for Bedrock

The post How Pfizer is Saving Lives with AWS’s Generative AI Services appeared first on Analytics India Magazine.

Image by Dall-E

In a world where technology is rapidly evolving, one topic stands out, capturing the imagination of scientists, tech enthusiasts, and the general public alike:

The rise of Artificial Intelligence.

As we stand before a new era, the question looms large:

What does the future hold for AI?

Stay with me and let’s discover it all together!

The AI Revolution

AI has already come a long way, from simple rule-based algorithms to deep learning models that mimic human cognition in solving problems and making decisions.

Image by Author

AI technologies are categorized into three main groups according to their ability to mimic human traits.

1. Artificial Narrow Intelligence (ANI) or AI with specialized abilities

ANI often known as weak AI or narrow AI, focuses on specific applications or tasks. It is designed to execute single tasks and tries to imitate human actions within a confined range of variables, limits, and scenarios.

Examples of ANI are prevalent in technologies like Siri’s speech and language processing on iPhones or the visual recognition capabilities in autonomous vehicles.

2. Artificial General Intelligence (AGI) or AI equal to human-level

AGI also known as strong AI or deep AI, refers to machines’ capability to understand, learn, and use intelligence to address complex issues similarly to humans. AGI operates on a ‘theory of mind’ framework, enabling it to perceive emotions, beliefs, and thought processes in other intelligent entities.

AGI remains a theoretical concept but has garnered significant interest from major tech firms. Microsoft, for instance, invested $1 billion in AGI through OpenAI. Today we already have ChatGPT-4, with its ability to tackle a wide range of problems and demonstrate higher-level cognitive skills, representing an early form of AGI.

3. Artificial Superintelligence (ASI) or AI exceeding human intellectual capacity

ASI represents a form of AI that exceeds human intellect, capable of outperforming humans in every task. ASI is not just adept at comprehending human emotions and experiences, it is also envisioned to possess its own emotions, beliefs, and desires.

While ASI is currently a theoretical concept, its anticipated decision-making and problem-solving skills are projected to be vastly superior to human abilities.

Before deep-diving into this super intelligence, let’s try to remember a bit…

What did the future of AI look like in the early past?

The concept of AI has oscillated between fear and fascination for years, predating the actual term. The prevailing belief was that true AI had to mirror human forms, obscuring the reality that AI had been operational for quite some time.

Notable achievements, like surpassing human skills in games such as chess, were just the tip of the iceberg. Since the 1980s, AI has been a key component in various industries.

Garry Kasparov playing against Deep Blue, the chess-playing computer.

The 1990s witnessed a transformation in machine learning with the advent of probabilistic and Bayesian methods. These advancements laid the groundwork for some of today’s most prevalent AI applications, including the ability to navigate massive data sets.

This capability extended to performing semantic analysis of raw text, enabling web users to effortlessly locate desired information among billions of web pages by entering simple queries.

The Quest for Super-Intelligence

Super-intelligence isn’t just about crunching numbers at lightning speed. It’s about a holistic enhancement in every facet of intelligence, from reasoning and creativity to self-improvement.

Imagine a world where machines innovate, think, and learn at levels beyond human capabilities. We are not still in such a world, but as our technology evolves, this scenario might be closer than we think.

Recent advancements, such as OpenAI’s GPT-4, showcase the rapid progress in AI. Considering all the breakthroughs experienced in fields such as machine learning or quantum computing, the emergence of super-intelligence is becoming increasingly plausible.

And this brings us to…

The Potential and Perils of Super-Intelligence

The benefits of super-intelligence are boundless. From the medical field with AI-based disease predictors to finances or climate change, a hypothetical super intelligence could enhance human society. However, the actual AI level is already causing some big impacts, so a superintelligence could even worse these scenarios:

1. Transforming the Workplace

Forget the old idea that AI will only affect low-skilled jobs. Thanks to advancements in generative AI, like DALL-E and Mid-Journey, even creative professions are feeling the heat.

These AI systems can churn out art, literature, and videos in a flash. They’re so fast that they’re starting to outpace human journalists in writing basic news articles.

This shift is raising big questions about the future of jobs, especially in fields we once thought were safe from automation.

2. Navigating the Intellectual Property Maze

The rise of AI is stirring up a storm in the world of intellectual property. When an AI creates a song or a logo, who owns it?

• The programmer?
• The AI itself?
• The creators who provided the training data?

This issue is getting more complicated as AI systems, trained on existing content, are now capable of producing incredibly convincing fake content. This dilemma has even led to legal showdowns, like Getty Images suing Stability AI over photo usage.

3. The Misinformation Challenge

AI’s ability to create realistic, fake content cheaply and quickly is a double-edged sword. This technology could massively amplify the spread of misinformation online, a concern that’s growing as fake content becomes more sophisticated.

4. AI in Decision-Making

Governments and businesses are increasingly leaning on AI for decisions in areas like social welfare and law enforcement. These systems assign risk scores that can hugely impact people’s lives.

However, there’s a catch: unchecked AI can replicate and even worsen existing societal biases.

Humans must stay in the loop, guiding AI decisions to prevent unfair outcomes.

Preparing for the AI Era

With great power comes great responsibility. As AI continues its rapid advancement, we need to keep up. Policymakers, industry experts, and developers need to collaborate on rules and regulations for AI.

Ensuring that these intelligent systems align with human values and ethics is paramount. Without proper checks and balances, unchecked AI could lead to dystopian outcomes, with machines potentially dominating humanity. The clock is ticking for decision-makers to craft policies that keep pace with this evolving technology.

Moreover, the equitable use and distribution of AI is a pressing concern. Super-intelligent AI could confer immense power to those who control it, leading to disparities in wealth and influence. Ensuring the beneficial and equitable use of AI is a challenge that society must address head-on.

This brings us to the…

Singularity theory

The Singularity theory was first coined by John von Neumann in 1958. For those unfamiliar with this concept, it describes a hypothetical moment when AI either develops self-awareness or gets such advanced capabilities that they evolve beyond human control.

Image by Craig Bellamy

In this scenario, AI would improve itself autonomously at an exponential rate, far beyond human comprehension or control.

However, this concept is highly controversial.

Critics against this theory argue that it underestimates the human mind while overestimating the potential capabilities of AI. And in case this event is to happen, the timing for such event is also a subject of much debate among scientists and technologists.

So let’s not panic just yet!

Navigating the Future with Optimism

The trajectory of AI’s development is promising. By adopting a balanced approach, we can harness the benefits of AI advancements while effectively addressing its challenges.

As we stand at this critical time in history, we must approach the dawn of this super-intelligence with a blend of excitement, caution, and responsibility.

How do we get ready for what’s coming?

The answer lies in raising awareness and continually educating ourselves. AI’s extraordinary capacity for automating routine tasks not only saves time but also allows humans to engage in more intricate and imaginative work.

Take healthcare, for instance: AI’s ability to interpret medical images can be life-saving. Similarly, in transportation, AI’s role is growing, evident in the popularity of self-driving cars like Teslas.

Future developments promise even more sophisticated automotive technologies. Moreover, AI is streamlining logistics and supply chains, enhancing efficiency and reducing costs.

Josep Ferrer is an analytics engineer from Barcelona. He graduated in physics engineering and is currently working in the Data Science field applied to human mobility. He is a part-time content creator focused on data science and technology. You can contact him on LinkedIn, Twitter or Medium.

More On This Topic

• The Rise of Vector Data
• GitHub Copilot and the Rise of AI Language Models in Programming Automation
• Why Emily Ekdahl chose co:rise to level up her job performance as a…
• Drag, Drop, Analyze: The Rise of No-Code Data Science
• The Rise and Fall of Prompt Engineering: Fad or Future?
• The Rise of ChatOps/LMOps

Jensen Huang is everywhere, and so is NVIDIA. The company seems to be stealing the spotlight at all the major events – be it Google Cloud Next, AWS re:Invent or Microsoft Ignite, bringing the party to life at each one of them.

During the recent re:Invent keynote, AWS and NVIDIA jointly announced a strategic initiative to provide a new class of supercomputing infrastructure, software, and services tailored specifically for generative AI.

The duo have also decided to deploy NVIDIA’s much-anticipated GH200 chips, initially slated for release in 2024. The installation of NVIDIA’s GH200 chips will occur within AWS’s cloud infrastructure, emphasising the global availability of this advanced hardware for AWS customers.

NVIDIA x AWS

The event also presented the initiative it has taken to set up the world’s fastest GPU-powered AI supercomputer, which will be a giant leap towards reshaping industries and driving technological progress at an unprecedented pace. This innovation is named Project Cieba and will aim to feature 16,384 NVIDIA GH200 superchips, which will process a staggering 65 exaflops of AI, propelling NVIDIA’s next wave of generative AI innovation.

At #AWSreInvent, @AWSCloud CEO Adam Selipsky and our CEO Jensen Huang spotlight the pivotal role of #generativeAI in cloud transformation, highlighting their companies’ growing partnership. https://t.co/TrmbOu3GXw

— NVIDIA (@nvidia) November 28, 2023

Apart from this, AWS is also working with NVIDIA on introducing three new Amazon EC2 instances, including P5e instances for large-scale generative AI and HPC workloads and G6 and G6e instances for a wide range of applications, such as AI fine-tuning, inference, and graphics.

“NVIDIA and AWS are collaborating across the entire computing stack, spanning AI infrastructure, acceleration libraries, foundation models, to generative AI services,” said the CEO of NVIDIA, Jensen Huang.

NVIDIA x Hyperscalers

Clearly, NVIDIA thrives in a collaborative environment, and at AWS re:Invent, it became clearer, as it also shares partnerships with rivals Google Cloud, Microsoft Azure and Oracle.

“Our partnership with NVIDIA spans every layer of the Copilot stack — from silicon to software — as we innovate together for this new age of AI,” said Microsoft chief Satya Nadella, at Ignite 2023.

At this event, NVIDIA and Microsoft announced their partnership to launch an AI foundry service on Microsoft Azure, aiming to boost the development of custom generative AI applications for enterprises and startups. This service integrates NVIDIA’s AI technologies and DGX Cloud AI supercomputing with Azure’s infrastructure, providing a comprehensive solution for creating and deploying tailored AI models.

The partnership also emphasised custom model development, leveraging NVIDIA’s AI Foundation Models and tools and making these advancements accessible through Azure’s cloud platform and marketplace. This collaboration signifies a major step in facilitating advanced AI application development and deployment in various industries.

“Many of Google’s products are built and served on NVIDIA GPUs, and many of our customers are seeking out NVIDIA accelerated computing to power efficient development of LLMs to advance generative AI,” shared Google Cloud chief Thomas Kurian at the Next event held mid-this year.

At Google Cloud Next, NVIDIA partnered with Google to drive advancements in AI computing, software, and services, alongside enhancing AI supercomputing capabilities.

The duo is working together to optimise Google’s PaxML for NVIDIA GPUs, facilitating large language model development, and integrating serverless Spark with NVIDIA GPUs for accelerated data processing. In addition to this, Google Cloud had said to feature NVIDIA H100 GPUs in its A3 VMs and Vertex AI platform and gain access to the NVIDIA DGX GH200 AI supercomputer and more.

Recently, Oracle also announced a multi-year partnership with NVIDIA to speed up the AI adoption for enterprises, which helps customers solve business challenges.

In a recent interview with AIM, Oracle said that it is well-equipped in terms of infrastructure, as NVIDIA selected OCI as the first hyper-scale cloud provider to offer NVIDIA DGX Cloud. “When NVIDIA thinks of cloud and data, they think of Oracle,” said Oracle’s Chris Chelliah. He said that it utilises MySQL HeatWave data for real-time anomaly detection on NVIDIA clusters for its customers.

NVIDIA is Omnipresent

NVIDIA’s diverse partnerships with leading cloud providers and hyperscalers uniquely position it across various facets of the AI landscape. While everyone is busy building their in-house silicon capabilities to handle AI workloads, the nature of partnerships seems to be changing rapidly.

From an innovation and AI advancements standpoint, Google Cloud seems to be NVIDIA’s favourite, while Microsoft Azure stands a pivotal partner for enterprise reach and application development, given its strong enterprise focus and extensive customer base.

Oracle differentiates itself in data management and AI-driven solutions, particularly through its emphasis on real-time data processing capabilities. AWS, on the other hand, plays a critical role in security-focused AI solutions, addressing the increasing concerns around AI security and reliability.

Overall, these partnerships provide NVIDIA with a multifaceted platform to expand its AI capabilities and market reach, with the impact of each partnership aligning with NVIDIA’s strategic focus areas, whether it be AI innovation, enterprise application, data management, or ensuring security in AI solutions. Simply put, everybody likes to NVIDIA.

The post Jensen Huang Brings re:Invent to Life appeared first on Analytics India Magazine.

There has been an enormous amount of media coverage about both side hustles and employees who want to continue working from home. The result, of course, is high demand for information on how to make money online. Selling ebooks has turned out to be one of the most profitable methods, even with many people hiring freelancers to do the actual writing. Now, however, you can eliminate those costs forever with a lifetime subscription to My AI eBook Creation Pro.

Digital products such as ebooks are profitable because not only do they not have the overhead associated with physical products, they never need to be removed to make room for new products. So you can just keep adding books to create a perpetual income stream.

The only problem is that writing books is not a fast and easy enterprise for most people. Fortunately, this happens to be a sector where the latest advances in artificial intelligence can have a major impact. My AI eBook Creation Pro uses ChatGPT AI to help you effortlessly create ebooks completely from scratch with just three clicks, even if you have no experience whatsoever in writing or design.

You simply provide the platform with information about your project, including name, category, topic, language, tone, target audience and maximum words per chapter. Using that data, a variety of chapter titles will be suggested, and you’ll simply edit each title and description. Then all that’s left for you to do is click on “Next” and AI will magically create the book.

That draft can be extensively customized, though, allowing you to modify any element of the book you’d like to change. This method is so much quicker and less expensive than hiring someone to write your books. All you need is a computer and an internet connection. If you’re not sure what to write about, you can read or listen to over 1,500 15-minute book summaries for inspiration.

Get a lifetime subscription to My AI eBook Creation Pro during our extended Cyber Monday Sale through December 3, while it’s available to new users for only $24.97, a $10 price drop from the regular $34.99 sale price.

Prices and availability are subject to change.