Autocorrect, a spell check feature, is pre-installed on most virtual key boards on Android and iOS phones. It also identifies misspelt words on Apps such as MS Word and Google Docs. It corrects the spellings even while we are typing the word.

There is predictive typing feature which infers the word or words which will appear next in a sentence. There is autocomplete to predict the words which will complete the sentence.

All this appeared in the early 1990s when Internet was nascent to enable faster and error-free typing.

There is a large market for the writing enhancement software. There is an AI-tool called Grammarly, a grammar checking tool. After the US, India is its largest market for it. Proofread of Google competes with Grammarly. It is a collaborative tool for Google Workspace called Duet AI. It operates on subscription model. Grammarly offers a free tier in addition to subscription plans.

ChatGPT generated material is subjected by the students to AI-paraphrasing tools such as QuillBot to avoid detection of copy lifting.

Previously, people were judged on the basis of the language used — grammar and spellings. By using software more and more people make sure that what they write stands out.

Audio-visual media these days teaches us new words. Previously we learnt new words from text and there was likelihood of getting its pronunciation wrong. Video teaches us the correct phonetic aspects of the new words. The earlier generations had good writing skills, whereas the new generation has good speaking skills. Even people love to write the way they speak.

LLMs can now be trained to understand how the new generation uses the language.

All of us are now aware of generative AI where the model is trained on massive data so that it is able to generate derivative of the data, either as a summary or something else.

Another type of AI is Predictive AI which also uses lot of data and subjects it to statistical analysis such as clustering and regression to predict an outcome.

NetApp works in this area. They have been working on it since 2018 in collaboration with Nvidia.

They use predictive analysis to predict failures in all types of systems. Clinical trial data of drugs can be compared to other data sets. This way they detect anomalies. It quickens the presentation of data to the regulatory authorities. Predictive AI can be used to predict forest fires. It can also be used in medical imaging.

NetApp also helps customers to store their images and documents. Google and Nvidia’s generative AI later can be used to search these documents, and avoid some documents being searched to let them remain private.

NetApp assists AI models to do ‘model traceability’ — keeping track of documents and datasets used for individual models. Models then can be compared in terms of accuracy.

It is difficult to monetize generative AI. However, predictive AI is easily monetizable, as its outcome is powerful and impactful.

We have already discussed the Google case of anti-competitive practices to maintain its dominant position. As we have already observed, the Justice department’s case against Google refers to a series of contracts where Google pays web browsers and smartphone makers to be the default search engine.

Michael Roszak, a senior Google executive (vice-president for finance) wrote notes on communications for a training programme in 2017. In the notes, Michael writes that search advertising is one of the greatest business models ever created. This business, he continues, ignores one of the fundamental laws of economics. It ignores the demand side of the equation (users and queries) and focuses only on the supply side of advertisers. He goes on to compare this search advertising business to illicit business of cigarettes and drugs.

The document was used as a piece of evidence in the case. Roszac testified at the trial in September, 2023. However, the government removed from the web public access to emails, chats and internal presentations at the instance of Google. The exhibits were reposted after the judge brokered a compromise to create procedure for their posting. Thereafter, Roszak’s notes were made publicly available on 28th Sept, 2023.

The document is full of exaggeration and hyperbole. Roszak testified that he could not recall any presentation on the subject. He further said the document was never sent to anyone else at Google. Roszak said he was saying things he did not believe as part of presentation in the course.

Hank Asher pioneered the data broker industry. In a technical field, he is not a known figure. However, his legacy continues to affect us, and has as much impact as any other Big Tech name from Silicon Valley.

Data brokers collect massive amounts of personal data. They cull data from public records, card transactions, social media and geolocation data. All the data is then synthesized for their clients and is used for advertising, insurance and law enforcement.

In the early 1990s, Asher’s partner had the idea of buying bulk data from Department of Motor Vehicle (DMV) from the state of Florida. The cost was a penny for a record.

They relied on a processing unit operating sequentially. However, they connected multiple smaller devices to distribute processing tasks. The system ran faster than that of the competitors.

Asher’s firm provided current address, past occupants of the same address, past addresses, past residential addresses, registered businesses against a particular name. More and more search criteria were added — marriage and divorce records, credit reports, gun licences, voter registrations. Home internet proliferated in the 1990s — email addresses and online shopping behaviour were added.

The users were the police department, insurers, law firms. DBT went public in 1966. Asher had 36 percent stake in it. It was worth $111 million. DBT was converted into ChoicePoint commanding an annual revenue close to $1billion (2004).

In 2000, during the presidential election DBT Online was roped in by the state of Florida to clean voter rolls by removing felons. DBT’s methodology was so flawed that it barred black voters from voting. There was a high margin of error. The election was swung to AI Gore. Asher denied any responsibility and loomed large in data mining.

It later prepared a list of potential plane hijackers.

Asher evaded charges of his role as drug smuggling pilot between the US and Central America in the 1980s.

Asher was roped in to identify missing and exploited children. Hank Asher was a colourful character. He expired in 2013. His Hank Show succeeded in demonstrating how credit bureaus are worse than even social media.

Our data is aggregated and is sold to enforcement agencies, immigration, and hospitals. Aggregated data does matter. It is of grave concern

Facebook and Google revenues is close to a half trillion dollars on account of targeted advertising on the basis of aggregated data.

Asher’s original firm was sold to LexisNexis and his last company TLO was sold after his death to credit bureau TransUnion.

LexisNexis and TLOxp are routinely used to find the phone number, collect information of past criminal record, locate relatives and neighbours of someone in the news.

In the 1990s, traditional programming languages C and C++ were used for coding. They were low-level or hardware-level programming languages. These help in developing logical thinking. The server operating system Linux was written in C. These languages are still being used to interact with the hardware.

Later, the applications became complex. They required many more lines of code. To do so, new languages were used — they were much easier for coding purposes. The low-level programming languages were mainly procedural to execute functions. They had minimum reusability. The new age languages have in-built functions for most applications. These languages are Python, Java and JavaScript. They came into existence in the 1990s. The open source communities constantly augment them. There are several libraries to support these languages.

In the era of AI and automation, new age languages such as Python and Flutter are used to support applications like IoT, big data or cloud computing.

The foundation of Android development is Kotlin which is based on Java.

A programmer today has to use multiple languages so as to align with the business problem on hand. Spring Boot is more aligned to microservices and cloud native architecture. There are performance and security concerns in blockchain, and hence Rust is the ideal language.

It is necessary to develop deep understanding of one core programming language. Then you can learn other programming languages necessary to accomplish a task.

At G20 meet, on Sept 7, 2023, a synthesis paper on cryptocurrency has been placed. It has put forth a comprehensive road map and building blocks for each country. Finance deputies have discussed the need for a template to regulate crypto assets, including a way forward to put into effect the IMF, Financial Stability Board (FSB) and a standards setting body.

A single country in isolation cannot handle the issue of cryptos effectively because of the pull and push of technology. It has implications on macro-economic stability.

All this will pave the way for global regulatory framework for cryptos.

The IMF synthesis and FSB paper says cryptos should not be granted official currency or legal tender status. However, the report argues that a blanket ban is not easy — technically demanding to enforce and hence costly. Central Banks should avoid holding cryptos as official reserve as they pose a risk to monetary and financial stability. It stressed unambiguous tax treatment of cryptos.

Policy makers should guard against excessive capital outflow. Cryptos pose risks to emerging and developing economies.

Credible institutional frameworks and comprehensive regulation and oversight are the first line of defence against macro-economic and financial risks posed by the cryptos. The countries should adopt Financial Action Task Force (FATF) standards.

Post FTX crash, it was expected that the days of cryptos are numbered. However, the scam has not proved to be the death knell of cryptos. It means cryptos have now become a permanent fixture in the global financial market. Cryptos have the potential for profit, and that is too great and too easy to attain.

There could be regulatory hurdles, but these will be resolved over a period of time.

Many fringe cryptos have become irrelevant. However, Bitcoin has survived, and showing an upward trend in value. It lures the newcomers. Bitcoin is an example of speculative trading. The more volatile the price fluctuations, the more exciting is speculation.

Ordinary people drive the crypto phenomenon. Here they have a yearning that the change agent that technology is, it can change money market too. Cryptomania is based on this thinking. However, rules of money market and investment do not change. Human nature falls prey to greed, and there is manipulation of this greed by others. This does not change.

It is a fact that more and more marketers and marketing managers are engaging with generative AI. Media effectiveness is enhanced by generative AI. AI affects three major media trends — advertising is a field where creative people thrive. AI enhances human creativity through its use. In copywriting, image production and content. Secondly AI causes the rise in generative search. Lastly. AI brings about generative optimisation — simplifies production, targeting and effectiveness.

There is a rise of chatbots using AI to promote brands and elevate customer experience.

Marketers must be cautious about using AI-generated content which may violate the IP rights.

High tech platforms will witness cut-throat competition. There are look-alike applications. Platforms will be protective about their data. Next year, the third party cookies approach will end. In the offerings, there will be uniformity. Brands will have to invest in attention to stand out.

Advertising on connected TV is forecast to experience a spectacular growth.

Technology advances and causes changes in job profiles. Some old jobs fade away, say tongawallas who operated horse carriages went into oblivion after motor cars appeared. Some new jobs appear, say radio cab drivers and then Ola and Uber drivers. There is replacement of some jobs — stenographers have gone and are replaced by computers and transcriptionists. Over the past few decades, robots, computers and software have taken over the work being done by human beings. Some work is being done more accurately and at lower cost.

With the advent of machine learning (ML), artificial intelligence (AI) and generative AI, there is replacement of jobs in office settings and on the shop floor in factories. In medical diagnostics, computer systems do the pathological tests faster and quicker. In radiology, X-raying and scanning are better interpreted by AI-assisted computers. The edutech sector has affected teachers.

ChatGPT answers queries, and can be used in customer service. Coding can be done by generative AI, instead of a programmer.

In creative fields, AI can be used to do editing and summarization. In advertising, different copies can be generated for promoting a product and different headlines too. Journalism can use AI to summarize a news story and to edit it.

When new jobs appear, there will be change in the organisation structure. There would be changes in job descriptions. Engineers can imagine new products, and use computer software to build a prototype. Such imaginative engineers will be called imagineers. Creators can do the product management functions of product development, branding and marketing. CEOs will do more co-ordinating work, rather than bossing over the specialists. A CEO will act as an orchestra conductor.

Educational institutions must rise to the occasion, and develop new courses and teaching methods to train man-machine combination of jobs.

AI, especially generative AI, can catalyze nation’s growth, and is acknowledged as vital not only for the companies but also for the government. Nasscom report has recommended that India needs sovereign GPU Cloud. As we know, generative AI models are neural networks with parallel processing capabilities. The traditional CPUs have been used by the data centres and not the Graphics Processing Units (GPUs). As GPUs are expensive, the shift from CPUs to GPUs is going to be expensive. The GPU making Nvidia produces A100 and H100 chips which are most sought after all over the world. India can be a world leader if it makes plans to procure these chips in sufficient quantities.

SoftBank together with Japanese government, Israel with government backup, UAE with government backup drive the generative AI projects.

India needs LLMs trained in Indian languages. That will bring the benefits of generative AI to the common citizens. The government has to unlock local data to train AI systems. Generative AI can be used to map the rural terrain for irrigation purposes. LLMs can be trained on satellite images.

Generative AI can be used to develop drugs for needs of Indian population. Indian pharma can develop original therapies.