Internet: Promises and Pitfalls

More than half of the world’s population (5.5 billion people) is on the internet. Of course, there are glitches and disruptions, but the most visited sites are obvious. Among the most visited sites are the search engines such as Yahoo and Google, social media platforms such as Facebook and X and chatbots such as ChatGPT. India too follows the global pattern and is fond of WhatsApp. There is a deeper story about the exceptions from internet in the dry list of websites. It shows that perhaps internet has not lived up to its promise.

The World Wide Web had its golden age. Internet promised to be an equalizer. It was later that misinformation became prevalent. Knowledge on internet was accessible to all — there were no gatekeepers. A villager could access the lectures from the hallowed lecture halls of the best universities. In fact, knowledge was just a click away.

However, internet has lost its halo. There are no barriers but that has allowed falsity to creep in. It is indistinguishable from truth. The social media could connect people, but they divide people too. The algorithms are content agnostic. There is AI revolution in the offing — it has brought in anxieties about the future of work, art and creativity.

Despite the gloom, everything is not lost. You can still find a fan of a literary figure if you look hard enough.

Vibe Coding

A workplace has emotional tone or vibe. This has to be gauged using technology and data analytics, thus offering actionable insights that bridge the gap between organizational goals and human emotions. This is the core of vibe coding. The underlying idea is that AI translates human intent into working code through ambient inputs, natural language and contextual understanding.

In tech circles, vibe coding is the latest buzzword. It has caught the imagination of software developers in the Silicon Valley in San Francico Bay Area of California as well as the IT hubs of Bangalore and Hyderabad.

The word has been coined by Andrej Karpathy, OpenAI co-founder in February 2025. It is an AI-driven approach to software making. It accelerates projects and lowers the barrier to creating software. The coders can focus on the creative aspects of development rather than being bogged down in technical details.

Imagine giving voice inputs to your computer in plain English and getting an app or software built in just a few minutes. Your application comes to life before your eyes. Vibe coding is the latest way for both the developers and citizen coders to build apps with AI taking their natural language prompts and going from idea to running an app without the need to edit the code manually.

Vibe coding has been made possible by the convergence of several factors. AI’s ability to write the code has advanced dramatically (since late 2022). There are advanced LLMs such as Anthropic’s Claude 3.7 and Google’s Gemini 2.5 Pro. It is a major shift. A developer acts as a pilot giving instructions and AI handles execution. These models can write code, generate documentation, create test cases and enhance efficiency. These tasks took days. These can now be completed in minutes.

Engineers will have to reskill and upskill. They could remain relevant previously by writing boilerplate code. Now LLMs can generate thousands of lines of such code in minutes. Developers should shift solving complex problems involving layered thinking.

Custom algorithms and non-standard workflows will still require human inputs.

At its core, coding is problem solving. Fundamental coding skills are essential. Even if AI advances, the developers will have to set the direction, provide the context and make the final decision. Developers will have to guide AI to review its output.

Vibe should be separated from velocity. AI tools are powerful but not infallible. There should be guardrails. The code can be beautiful but functionally fragile. There should be testing, reviews and governance.

Tamasha

Tamasha troupes are booked to perform almost all days in Chaitra and Vaishakh. Only 10 free days are left. The mood during this period is upbeat in villages and residents want entertainment and are ready to pay for it.

The groups congregate at Narayangaon. They are from neighboring districts and all parts of Maharashtra. They pitch their tents and welcome village leaders who come here to book the shows. There is fair in Narayangaon — an annual ritual. It is a meeting ground for tamasha groups and buyers (say a village sarpanch or mukhiya).

The tamasha groups are named after their founder. Almost 30 groups come for the fair. This year (2025) is good for sales — the group is paid Rs. 3 lac for an evening’s programme. The team managers represent the troupes. The combined earnings of 30 groups are around Rs.12 crore.

An average tamasha group has 40 artistes. There are as many supporting staff — cooks, technicians, drivers of vehicles. These troupes take loans from private money lenders around May and repay them the next April. The loans sustain them to the next season.

Previously, there used to be a full tamasha programme beginning with Gan Gavlan — an ode to Ganesha and Krishna Leela to Vagnatya — featured plays. There were social messages about alcoholism and female foeticide. These days the performances consist of Hindi-Marathi music, dances and orchestra. The audience for Natak is dwindling. It used to start late at night.

Previously, tamasha was brought to the village for yatra of Gram Devta. That has not changed.

It will be difficult to fund a tamasha programme. There were no alternatives for earlier generation. These days we have cell phones and TV.

People leave as soon as the Natak starts. Still the payment remains — Rs.2- Rs. 2.5 lacs per show.

Many troupes do not perform for seven months. The performing days are reduced to 40-50 days of Chaitra and Vaishakh. Formerly, the audience consisted of 3000 people. These days there is a crowd of 300 people. The salaries of artistes have gone up to Rs. 2000 -Rs 2500 per show. Banks do not finance the shows.

The established groups are putting up a fight. They get repeated booking since the audience loves them. There should be a supportive scheme for tamasha artistes providing them a minimum 100-days performance at state events among others.

Pocket FM

Pocket FM is audio platform launched in 2018, which provides a TV-show like content in audio. The episodes are 10-minutes byte-sized and offer serialized fiction. It has all the elements of cinematic experience such as voice acting, sound effects and a background score.

In 2020, the consumption time per user per day has increased from 30 minutes to 120 minutes. The first viral show was Kitni Mohabbat Hai commanded a million downloads.

They experiment with different pivots to find the product-market fit. They got investors after reaching a target of 1000 users. It has now 200 million users globally and has successfully expanded into the US market. Currently, 73 per cent is their revenue come from the US and 5 per cent from Europe. Their seven audio series have joined the Rs.1000-crore club.

Their aim is to empower creators to reach their audiences without gatekeepers.

Ad Revenues on Q-Commerce Platforms

Millions of consumers use q-commerce platforms to order essentials, fashion, cosmetics, kitchen appliances. They can now monetize their large user base through targeted advertising.

These platforms are now generating an estimated Rs.3000-3500 crore annual recurring revenue (ARR) from advertisements.

Blinkit leads with a lion’s share of 45 per cent, followed by Zepto 35 per cent and Swiggy Instagaram at 20 per cent. Amazon India’s ad revenue for 2024 stands at Rs 6700 crore (only twice as much), though q-commerce accounts for only 8 per cent of online shoppers.

The surge shows a fundamental shift in consumer behavior. Q-commerce started as a top-up channel. It is evolving into weekly and monthly purchases channel. This change gives a boost to advertising.

Q-commerce is affordable and high-impact way to build visibility and test products in the market. The biggest drives are impulse-buying products, e.g. ice cream packs, beverages.

The cost-per-click rates are on par with or even higher than Amazon’s.

Coding Automation

Across the startup ecosystem, coding automation is on a superfast track. The automation levels are to the extent of 15-50 per cent. The future target by the end of 2025 is between 40-85 per cent.

Coding automation has become a strategic priority. Startups use generative AI tools to reimagine the way the software is written, tested and deployed. These startups are e-commerce, fintech to SaaS startups. The engineering teams are freed up for higher-value innovation.

InMobi, ad tech firm, has automated 50 per cent of its software coding. It is targeting 80 per cent automation by the end of 2025. Udaan, e-commerce platform, automated 90 per cent front-end and 50 per cent backend systems. LeadSquared, SaaS unicorn, has integrated AI tools such as GitHub Copilot and Claude code to automate routine coding tasks. It embeds generative AI throughout its software development cycle. Gupshup has automated 35 per cent of its coding workflows. Co-Rover has reached 40 per cent automation.

The goal is not to replace human developers. Instead, automation is being deployed to eliminate repetitive, low-value tasks so that engineers can focus on critical thinking, user experience, and system optimization.

Controlling AGI

DeepMind, Google has published a paper which predicts the arrival of artificial general intelligence (AGI) by 2030. This paper is a 145-page document and is co-authored by Stane Legg. It points out the risks that AI could pose. These risks are put into four categories — misuse, misalignment, mistakes and structural risks. The first two risks are elaborated in the paper, while the last two risks have not been discussed in detail.

Misuse may refer to a user asking the model to fabricate a virus. The developers must develop safety protocols to prevent this. The system’s capabilities should be restricted to comply with such misuse.

If the system pursues a goal that is difference from human intentions, it is called misalignment. It is just like terminator movie. You ask the model to book the movie tickets, and it hacks into the ticketing system. It surpasses the safety measures. This is misalignment.

Mistakes can be mitigated. It is to be seen how they can be mitigated. The system should not be made very powerful. It should be deployed slowly.

Structural risks refer to the consequences of multi-agent system which spew out false information that sounds so convincing.

We should discuss potential ways AGI could harm beings before deploying it.

US Tariff Disruption

Global trade order has been disrupted by the US tariff assault. America was committed to free markets, and the irony is that it is disrupting the trade order. Of course, it has been triggered by a widening trade deficit with the rest of the world — yearly $ 1.2 trillion. It has devastated the industrial base of the US in the last half a century. However, the additional tariffs are not confined to the countries which run trade deficits. There is no positive discrimination. It is a protectionist policy for the US industry. It could do damage to the US industry as well. US could face recession over the next year or so. It could face stagflation.

The tariffs seem to have been computed by an obscure formula linked to trade deficits. Trump clarifies that its tariff could be contained if countries take steps to increase market access for the US firms. There could be tariff wars and a reorganization of world trade order.

India faces 27 per cent tariff on its exports to the US. It will be added to the extant average US tariff at 3.3 per cent. Product-wise duties on Indian exports could be higher in a few cases. The extra imposts on India are lower than many at India’s key Asian competitors. It is comforting for India that pharmaceuticals and semiconductors are exempt from reciprocal levies.

India now must negotiate with US bilateral trade agreement in a mutually beneficial manner. Along with this, there should be domestic reforms to boost productivity. India in short-term may lose in labor-intensive sectors such as sea road, gems and jewellery . However, India could gain in textiles which is equally labour intensive. India could have advantage in electronics. The US is not in favour of non-tariff barriers to trade. India must exercise caution here. India can consider legitimate concerns of domestic industry, but these should not come in the way of business practices.

LeCun and AI

LeCun was born in France in 1960 and has been fascinated with AI since an early age. At the age of 9, he saw A Space Odyssey and that made him fond of space travel, AI and human intelligence. He learnt at young age that intelligence is self-organizing — complex behavior emerges from the interactions of simple elements.

LeCun started his professional work in 1980s. Neural networks then had fallen out of favour. There were limitations of ‘perceptrons’, some of the early neural networks, first pioneered in 1950s.

The field of AI shifted to symbolic and rule-based systems. The field was revived by people not obsessed by history. They established a connection between statical physics, theoretical neuroscience and neural nets.

He did his PhD work in Pierre University in mid-1980s.

He developed the renowned backpropagation algorithm — his first contribution to neural networks. Here the network learns based upon errors detected in its output. These are ‘backpropagated’ through the network to adjust internals weights. It improves accuracy. It has now become an established technique to train the network. He completed his doctorate in 1987.

He joined the Univrsity of Toronto for post-doctoral fellowship under Geoffrey Hinton. A year later he joined Bell Labs. He contributed to the development of CNN — convolutional neural networks. CNNs scan images. They detect features like edges, textures and shapes. They detect these irrespective of where they are in the visual field. It improves computer vision (CV). It revolutionized handwriting recognition, cheque clearing and facial recognition. It helped medical imaging, autonomous vehicle perception and AR.

He also had stints at AT&T and NEC. LeCun later joined New York Univrsity in 2003. He still serves as Silver Professor. He was recruited by Zuckerberg for Facebook in 2013 and is currently Chief AI Scientist.

LLMs have limitations. They are just token generators and do compute of fixed amount to generate a token. It is a reactive system. It is an intuitive system. The slower, deliberate reasoning system of human brain is another system. It is easier to teach AI systems higher order skills, say chess playing and clearing tests — cortical based reasoning skills come much later. These require more cognitive effort. This is called Moravec’s paradox. The machines act exactly in reverse direction. LLMs master higher order skills, say NLP. But they lack foundational abilities. The real world is messy and continuous. Humans process more data than AI systems.

LeCun is pioneering an alternative approach Joint Embedding Predictive Architecture (JEPA) that mimic that physical world on visual output. The predictions do not happen in the space of raw sensory inputs. They happen in abstract representational space. In another V-JEPA or video JEPA model, LeCun’s team trained a system to complete partially occluded videos.

LLMs are probabilistic while predicting next token. LeCun’s system learn to represent the world at multiple levels of abstraction and predicts how these representations evolve under different conditions.

Larry Ellison, the Media Magnate

Larry Ellison co-founded Oracle and has a net worth of $175 billion. He has built a Japanese style imperial villa costing $ 200 million near Palo Alto, California. He owns the largest Hawaiian island. He has married and divorced several times.

He purchased a 22-acre estate in Florida near Palm Beach. The price he paid was $173 million. He invested in Twitter too when Musk bought it.

He is now 80. He surrounds himself with beautiful things. He is expanding his corporate empire too. Oracle emerged as a possible bidder for TikTok, the video app of the Chinese company ByteDance. The company has either to divest or be banned in the US. It has an April deadline. Oracle has minority stake in TikTok. His son David bids for Hollywood studio Paramount.

Ellison is a part of AI infrastructure project Stargate.