In yester years, before the advent of ChatGPT or generative AI, or AI and LLM becoming mainstream, the ML models were built by data scientists in the employment of companies in areas like retail and banking. ML models were applied to the data. These models used to do everything — from prediction to recommendation.

After the arrival of LLMs, the model building work is being outsourced to external organisations. All the foundation models have been developed by OpenAI. Many new organisations are building LLMs for specific areas.

It does not mean that the work of data scientists has been replaced. or eliminated. However, it spares data scientists for other work, rather than spending energy for building a model.

Data centres which house complex matrix of servers in steel racks provide massive computing power to leverage AI applications. Apart from servers, what we require are optical fibres (OF), power supply cables, computer fans, CPUS, GPUs and routers

These computers generate heat and require cooling through layers of water pipes across the whole data centres and airconditioners.

The traditional data centres are designed with all this hardware, but for AI applications, we do require computing power manifold. These data centres require advanced computing capacity and power management. Existing hardware could be used for AI, but it is a lot slower. AI runs on huge clusters of power-dense GPUs.

AMD has introduced data centre accelerated processing unit (APU) for high pertormance computing (HPC). To handle LLMs, AMD introduced the M1250x accelerator. It is used in supercomputers. AWS focuses on hardware to deal with AI. It has developed AWS Trainium, a high performance ML chip.

LLM training has increased the requirements of storage at the data centres. There is a shift from CPU-centric traditional architecture to GPU or TPU-centric set ups capable of parallel processing.

Data centre now need more number of servers or servers of high density. Workloads going up to 40 KW

per rack are required to cater to AI.

There are hyperscale data centres to handle complex AI workloads. Liquid cooling systems are extensively used to reduce cabling and operational expenses.

AI is evollving by leaps and bounds. It requires hardware and physical infrastructure. At the same time it requires software management. Microsoft offers hyperscaler service and Microsoft Cloud. AI solutions are adopted by more functional areas of business. India is matching this trend.

1. Autonomous Cars :

Deep learning is widely used in driverless cars. Deep learning analyzes the real time data captured by the sensors and cameras so that the car navigates on its own. It prevents untoward events, such as collisions and pedestrian harassment.

2. Natural Language Processing (NLP) :

Deep learning makes the computers to understand and generate human language. Deep learning powers the chat bots. Deep learning makes computers understand the context and intent, summarize text, do question answers, and translations from one language to another.

3. Diagnostics and Imaging :

Deep learning facilitates diagnostics and imaging. Medical practitioners can identify diseases from the images. The computers detect the anomalies in the images. Here algorithms are used. These images could be the X-rays, CT scans or NRI scans. Diseases such as cancer, cardiac conditions and neurological disorders could be easily and precisely diagnosed.

4. E-commerce :

Deep learning is used by e-commerce platforms — customised products could be offered to the customers. There could be precise targeting. Consumer behaviour is better analyzed. The product offerings could be adapted to consumers preferences. It creates a pleasant shopping experience for the consumers.

5. Fraud Detection :

By analyzing financial transactions and detecting unusual patterns, financial frauds can be predicted .

6. Entertainment and Content Creation :

Deep learning is used for creating content and scripts. It also helps in generating music. AI can generate new material.

7. Agriculture :

Agriculture can become smarter and efficient. Drones can be used to monitor crop health. These drones can be AI-assisted. Yields can be increased and wastages reduced.

8. Manufacturing and Quality Control :

Deep learning identifies flaws in the products and helps in maintaining standards. AI-assisted inspections are used. IoT is used to have communications between machines in Industry 4.0. Wastages are reduced. AI is used in production planning and control (PPC)

9. Energy and Power Grid Optimization :

In energy sector, deep learning is used to analyze data captured by sensors to predict energy consumption and optimize distribution.

10. Drug Discovery and Development :

In the field of pharmaceuticals, deep learning speeds up new drug development. Potential molecules are identified by studying their structures. The chemical interactions could be is expedited.

Gen AI was restricted to IT departments to begin with. However, it has now diffused to other departments of the organisation such as production, HR, marketing, finance and supply chain. Thus generative AI is extending itself to many other functional areas of business.

AI works for all, and now with a chatbot interface, it has spread everywhere in the organisation. Just as internet spread everywhere after the introduction of the search engine in the late 1990s, generative AI too has become all pervasive after ChatGPT.

Even different businesses such as healthcare, banking, finance, insurance, telecom, retail have adopted generative AI.

Google unveiled new capabilities in Vertex AI platform for developers to build, test and train new AI apps.

Duet AI has capabilities such as coding assistance and database migration.

On 31st August, 2023, Google announced launching of its AI-powered search in India. It will give Search Generative Experience (SGE). Google Search gets a layer of generative AI. Users will have to opt for this. Traditional search shows links of multiple web-sites that have information related to search queries and key words. In AI-powered search, there would be an over-view of key information. It has also integrated Text-to-Speech feature.

In traditional search, the link may or may not have relevant information. AI-powered search, the relevant information is compiled, organised and shown upfront. The sources of information are indicated. There are photo results at the end.

Each search query is treated as a new command. One can ask follow-up questions or queries, and AI-assisted search would carry over the context from question to question. AI-assisted search also understands long questions, and the question need not be broken down into smaller queries.

It supports both English and Hindi languages. There is a language toggle button on the top left side of the information screen,

Google will soon roll out a feature that would allow users to make use of voice to ask follow-up questions (instead of typing them).

As generative AI is a nascent technology. it has some limitations. It is not completely accurate. Google restricts the number of queries. The AI search too may carry ads.

As we know, artificial intelligence systems are trained on vast amounts of data. Companies in this field rely on sub-contracted staff to deal with data. The staff is located domestically or abroad. They lack the benefits available to the regular staff on company’s pay-roll. It is rightly called ghost staff, and the work done by them — labelling data and rating responses — is ghost work.

There are millions of such data workers around the world. Their task is mundane and stressful. The wages are low. There is constant supervision. They are given scant training. All this is enough for bias sneaking into the AI system being developed.

Big Tech builds AI on the bulks of exploited workers.

US Lawmakers, led by Senator Markey has written to Big Tech firms a letter asking the companies about the working conditions of these workers.

There is an antitrust suit against Google, which as the Justice Department says, is because Google has abused its market power to bully trade, and protect its monopoly and thwart competition.

It sounds similar to Microsoft suit some 25 years back. In both the cases Big Tech has been accused of using its market power to cut competition unfairly.

Of course, the Microsoft suit received wider attention, and it is doubtful whether the Google trial will command the same attention.

Microsoft was a lone tech titan in the 1990s, and its Head Bill Gates was a celebrity. The trial started in 1998, and the testimony was recorded over eight months. It was widely covered by the news agencies. The proceeding received coverage in the NYT daily. Many important concepts were discussed during the trial — network effects and switching costs.

Microsoft’s defence was that the snippets of text were taken out of context, and were not evidence of its anticompetitive conduct. Lawyers from Justice Department and States were of the view that they have clinching evidence.

A federal judge found Microsoft guilty of violating antitrust laws. However, he was not in favour of the remedy put forward by the government — break the organisation.

In Google case. Justice Department refers to the Microsoft case and the tactics it adopted in 1990s. Google weilds its might in online search as Microsoft did in operating system (Windows).

Netscape, the pioneer of internet browsing, bore the burnt of Microsoft’s bundling of its Explorer browser with Windows. Since 90 per cent computers were loaded with Windows, it had the key to access Internet.

Google has to work with partners and pay them — smart phone companies, browsers and other devices.

Though Google’s tie-up with Apple and Samsung made it the default search engine, it is legal as it is in the best interests of the consumers. It reduces cost for device makers.

Google pays more than $10 billion a year to maintain its position as the default search engine on web browsers and mobile devices and thus stifles competition. This is what the US Justice department said at the start of the trial in Washington.

The government lawyer said that this is the case about the future of Internet. The evidence will show that there is default exclusivity to block the rivals.

Google controls 89 per cent of the online search market. The company can afford to pay billions for defaults as it is uniquely powerful.

Google has abused its position for the last 12 years by monopolising general search.

The first phase of the trial will assess whether google has illegally monopolised the online search market. The decision could be taken next year (2024) on whether Google broke the law. If the justice department succeeds, at the second phase of the trial, may seek remedies to break the search business from other products like Android and Google Maps.

Microsoft operated in the 1990s in world that flirted with internet. It was an early tide. E-commerce came later, and then there was a race to jump on the digital bandwagon.

Then came the mobile computing. Blackberry was used basically as an email device, and was introduced in 1999. The new millennium has witnessed the smart phone era since 2007. In the early days, one got online through a computer (PC) that ran on Windows Operating System. Microsoft entered into media business in 1996 and partnered with NBC.

Since then, Microsoft has stopped selling Windows. It has entered cloud computing. And now it has partnered with OpenAI to enter generative AI field. Microsoft Bing is now powered by ChatGPT.

Amazon and Flipkart are online market places. If a product is advertised on them, the chances of conversion are very high, as the product is at a place where the buyers are at a point of purchase. The return is between four times to eight times. Ad spend thus on e-commerce sites is increasing. In 2022, these e-commerce sites accounted for Rs.7000 crore of the total Rs.49,900 crore spent online by advertisers, amounting to 14 per cent. In 2021, this amount was 12 per cent. Retail media is the biggest growth story in advertising.

The growth can be attributed to searches that happen online for the products the buyers are interested in. These searches originate on retail media and lead to sales. There is sharpness of targeting.

Social media such as Facebook, Threads, Instagram, Reels and WhatsApp do attract advertising. They constantly innovate to attract more advertising, say Reels. Google Ads work best on YouTube after search. New platforms lead to better click-through rates and better conversations.

Amazon’s global advertising revenue is about $38 billion, which is less than Google’s $166 billion or Meta’s $112 billion. Amazon is rapidly growing, doubling its share. Google and Meta maintain their share, but are not growing. Even in Google’s ad revenues, a large part is contributed by search advertising.

In India, brands in the automobile, consumer goods and cell phones are shifting to online retail media platforms.

The CPM on retail media is 30 per cent.

These days industry is receiving STEM graduates well. STEM stands for Science, Technology, Engineering and Mathematics. All STEM professional’s are not on par. Alumni of elite premier institutes such as IITs, NITs, IISC are very well-received here in India, but they go abroad for better prospects.

The quality of STEM education in India is not uniform. Top tier institutions do maintain standards. Many institutes have poor standards, outdated curricula, shortage of faculty. Management of these institutes are interested in commercializing education.

India produces 15 lac engineering graduates a year, of which only 2.5 lacs gets worthwhile employment. As against this, US produces 70000 engineers annually. The starting salaries for engineers there is $50000-$70000 per annum. The salaries increase and keep pace with the cost of living. By 2020, the average salary of engineers in the US hovered around $70000-$100000 per annum. The salaries are higher in Silicon Valley. In India, the salaries for the last two decades are Rs.2.5 lac to Rs.3.5 lac per annum. It is sheer exploitation. Most of our engineers are underpaid and underemployed.

There are face search engines, say Clearview AI and PimEyes, which have the capability to pair photos from the public web. These tools are available to the police which can identify a snapshot of someone by comparing it with online photos where the face appears. The identification reveals the name, social media profiles and other information which perhaps the person would not reveal to the public, e.g. risque photos.

These are technological breakthroughs. And the breakthroughs are ethical.

Tech giants such as Facebook and Google had developed the face recognition technology years earlier but they had held back the technology, thinking that it is too dangerous putting a name to a stranger’s face.

As far back as 2017, at Menlo Park, California, the HQ of Facebook, Tommar Leyvand, an engineer demonstrated a facial recognition software. It identified with a mobile camera the face of Zach Howard, and Howard confirmed the identification is right. The phone then identified several people correctly. This technology is a god-send for a person with vision problems or face blindness. However, it was risky.

Facebook had previously deployed facial recognition technology to tag friends in photos, but there was a hue and cry about the privacy. It was in 2015. Facebook faced a lawsuit costing company $650 million.

The new facial recognition software of Leyvand could enable users to recall the name of a colleague at a party or search someone at a crowded place. Still, Facebook did not release the version. In the meantime, Levyland left Facebook and joined Apple to work on its glasses — Vision Pro AR.

As early as 2011, a Google engineer worked on facial recognition tool. Months later, Google Chairman Eric Schmidt declared that the technology has been withheld.

With recent releases of the startups, the taboo has been broken. Facial recognition technology has the potential to become ubiquitous.

It helps the police to solve crimes. Authoritarian governments use it for surveillance of their citizens. It can soon become an app on our phones, or in Augmented Reality (AR) glasses. We will usher in a world with no strangers.