Blog

Warren Buffett is stepping down as chief  executive officer of Berkshire Hathaway Inc. This company was founded by him along with Charlie Hunger and is operational for the past sixty years. Buffett is considered the greatest investor of all time.

He dismisses the idea that anyone has the ability to predict future. He attributes his success to patience and investing in business that he understands. His returns are about double the returns of the S & P 500 Index. They have delivered a 20 per cent compounded annual return between 1964 and 2024. The business is  now valued at $ 1.6 trillion plus, supports 390,000 employees  and has a cash balance of $ 347.7 billion.

Buffett gives credit to people around him including vice chairman (for non-insurance operatistons) Greg Abel and Munger. He believes in ‘quality’ buying great companies at the right price.

In the 2008 financial crisis, Buffet scooped up investment in Goldman Sachs, GE and Dow Chemical. Yet he dismisses the notion that he has great  foresight.

Buffet is extremely humble. After moving in New York for some time, he returned to Berkshire in Omaha, Nebraska. He was born there. He had purchased a house there in 1958, for $31500 and continues to stay there. He is nicknamed the Oracle of Omaha, a misnomer perhaps. He treats Berkshire shareholders as co-owners in the company. At the AGMi in Omaha, he sits for hours taking questions from the audience.

In recent years, his company has experienced some lag. Corporate valuations reached dizzy levels on account of cheap and easy money. Berkshire is keeping pace with the S & P 500 benchmark. However, the investing industry has changed. It has become more democratized.

Buffet’s personal net worth $168.6 billion. it is partly a result of a long career and magic of compounding.

He announced in May 2025 that his successor by the end of the year will be Able(62).

David Sokel resigned in 2011 and he was a service contender to succeed Buffett. He had apparently violated company policies by buying Lubrizol shares. Abel is low-key and has been cut from similar cloth as that  of Buffett. He took more public duties in recent years, including sitting on stage with Buffett at Berkshire annual meetings to answer investor’s questions.

AI has the potential to shape human destiny, and therefore we deserve to understand it. Anthropic CEO has just remarked that nobody really knows how AI works.

Dario Amodei, Anthropic CEO is puzzled by the enigmatic nature of AI. A summarisation of a document by generative AI does not give us any Idea as to how it chooses the sentences and certain words over others. And despite being accurate most of the times, why AI makes mistakes at times.

Those who build AI systems do not understand how their creations work. Of course,this is a matter of concern.

On one level, we can say it works on a huge dataset fed to it, and allows statistical systems to mine for patterns which can be reproduced. The entire effort is to get driven by ingested material.

In the history of technology, this lack of understanding is unprecedented.

The CEO of Anthropic and his sister Daniela left OpenAI in 2020 as they were concerned about the organisation’s safety practices. The Amodeis and some five more colleagues at OpenAI founded Anthropic in 2021. They want to build safer AI.

They want to figure out the inner workings of these AI systems.

Ever since ChatGPT was launched by OpenAI in November 2022, AI has been considered to be a transformative technology somewhat similar to Industrial Revolution (1760-1840). Some go to the extent of saying that superintellligent AI will take over humanity. However, AI researchers at Princeton y beg to differ — they contend that the transformation on account of AI will be a slow process.

AI breakthroughs do happen. That happens with every technology — say Internet. AI is a general purpose technology. Humans would exert control over it. It is based on model training. However, AI has learnt everything it can from internet. In future, AI will have to learn by interacting with people, doing experiments in the real world and by being deployed in the organisations. After enhancing its capability, there will be its adoption. It is called innovation diffusion feedback loop.

In past, electricity and internet have been adopted slowly, and took decades before capabilities translated into economic impact. Diffusion of general purpose technology is slow. It shows over decades.

The impact on society of AI be realised when it is employed in productive sectors of the economy. This requires formulation of suitable policies. Besides every adoption requires different guardrails. The government s cannot curb AI models and should show resilience.

OpenAI was founded as a non-profit company and later became a capped profit company. In a normal capital structure company, investors can get unlimited returns. This stimulates economic activity in Silicon Valley. OpenAI’s returns have been capped at 100x. It is in the race to build AGI that will create trillions of dollars new wealth. Musk has sued OpenAI for becoming a for-profit company, but the judge has rejected the request to stop OpenAI but has allowed other parts of the lawsuit to go ahead. OpenAI has deviated from its original non-profit mission by being capped profit company and has now lifted the cap on investors so as to allure investors to maximize profits. It will be difficult to prioritize benefitting all of humanity over shareholder returns. It is likely to be tempted to create AI systems quickly without due caution for safety, security and fairness that such technology deserves.

As of now, OpenAI will ‘continue to be overseen and controlled’ by its non-profit Board. OpenAI wants to have its money both ways.

Teleportation is considered science fiction — travelling at great speed to the next location. This fiction has become a reality between two supercomputers. It is one of the holy grails of quantum physics. The quantum information is teleported. At the core of quantum teleportation is the concept of entanglement. Here two particles are so close to each other, despite their being away from each other. The state of one influences the other particle immediately. Einstein called this ‘spooky action at a distance’. It is not seen, but it can be felt. The effect could be witnessed. Spooky means ghost. Entanglement is used to teleport the state of a single qubit (the quantum equivalent of a bit). The teleportation is from one quantum processer to another.

In classical computers, the information is transferred by physical travel through wires or fibers. Entanglement does this — it reconstructs information on the other end. It is a leap for mankind.

Quantum communication can transform what is already known. Transfer of huge data quickly through machines all across the world. This is what could be achieved through qubits teleportation between supercomputers.

Ions were manipulated by Oxford scientists in electromagentic fields. The teleportation process works out depending on the control that entaglement has over the states of ions.

It is a milestone development for information sharing and processing. It could give birth to quantum internet. It breaks global barriers.

It is prudent on the part of UAE to have tech fuelled ambitions by leading the world in AI, while seeking to pivot theire economy from fossil fuels.

As they nurture technological ambitions, Dubai, the desert down known for its shiny high-rises, fast life plans to become the first nation in the world to deploy AI to draft new laws and amend existing laws to accelerate its law-making process. The AI-driven regulation system will be overseen by the newly created Regulatory Intelligence Office. The system has been announced recently and will use a database of federal laws, court rulings, government services, and will provide legal updates.

The role of AI goes just beyond drafting. The system will track the real-world impact of the new laws on Emirates’ population and economy. It will enable the administration to adapt legislation in real time.

There are issues about the propensity of AI models to produce unreliable outputs, hallucinations, potential shortcomings with legal interpretations.

Despite the risks, UAE has made AI a national priority. It is committing heavy investment in the sector through its wealth vehicle MGX. The scale of UAE and its neighbour Saudi Arabia is the focus. The UAE’s centralised government structure facilities this kind of technological experimentation. This is difficult to achieve in other countries having decentralised political structure.

Researchers from Oxford University have introduced a new algorithm for training neural networks that eliminates the need for both the traditional forward pass and backpropagation. It is called Mono-forward algorithm — a new approach to neural training by leveraging local errors for learning. This new algorithm is inspired by Geoffrey Hinton’s Forward-Forward framework. Conventional training relies on backpropagation. Mono-forward optimizes each layer using locally available information. It removes dependence on global error signals. It enhances biological plausibility and computational efficiency.

The researchers examined Mono-forward on multi-layer perceptrons and CNNs across benchmarks such as MINIST, Fashion- MINIST, CIFAR-10 and CIFAR-100. The results match or surpass the accuracy 0f traditional backpropagation methods. In addition, it provides reduced or more uniform memory usage, improved parallelization and comparable convergence rate.

It has improved efficiency of neural network training methods. This approach could pave the way for advances in neuromorphic computing and energy-efficient hardware. One can refer to the original paper: Mono-forward — Backpropagation Free Algorithm for Efficient Neural Network Training Harnessing Local Errors.

OpenAI has launched a new shopping option on ChatGPT. It answers a user’s query about the products by presenting him a choice from an array of products. A user’s query declares his shopping intent. ChatGPT displays the relevant products in visually rich carousels. It provides additional details and links to users to websites where either they can get more information or can purchase the products. This feature is available on GPT4o and 4o mini. It will be added to Plus, Pro, Free and logged out users. At present, the information is derived from third-party providers. It is exploring the option of providing information directly from vendors.

Could this feature be monetized? Currently, OpenAI says that the showcased products are not ads. However, will this be maintained forever? Could AI chatbots become the marketing platforms in future? In other words, they could become another online classified platform, undermining the productivity narrative being propagated by chatbot companies.

A chatbot presents a product in carousel when it perceives it relevant to user’s intent. It asseses the intent on the user’s query considering the context such as memories or custom instructions. It may consider the likes and dislikes of the users.

Google too has started using AI to show users products relevant to them. ChatGPT’s feature resembles this Google initiative. Google shows the products in listings in search. At present ChatGPT searches are not paid placements, but organic results.

Vibe coding is making waves. It is a term put in circulation by Andrej Karpathy in February 2025 on a post on X. It is a new type of coding where you follow the vibes, embrace exponentials and forget that the code even exists. The instructions are given to AI tool which does the coding for you.

The tools utilized for vibe coding are ChatGPT, Cursor, Replit and Bolt. These are popular. There are others such as Claude, Copilot, Lovable and Windsurf. Google too launched Firebase Studio in early April 2025. It is a cloud-based AI tool that builds apps simply using your prompts.

These tools assist engineers in coding. The workload that needed 4-5 engineers is now being handled by 1-2 at the most. This has improved efficiency. It affects fresher hiring in engineering. It could spell a death knell for coding jobs. IT sector is already under stress. The openings are shrinking. In India the enrolment for engineering courses is dropping by a substantial percentage. Partly, it could be attributed to the rise of AI. Developers will have to pick up additional skills.

AI understands 16-17 legacy codebases being used to interact with each other. Its results are not optimal. It reduces the workload. We still need senior engineers to make architectural decisions, to think beyond scale and to take ownership of the modules.

Learning programming is essential. The AI tools are not robust or built for full, production-level products.

In future, we may have better versions of vibe coding.

–

As for many other consumer products, there is a reluctance on the part of consumers in the US to spend $6 on a latte. We know the prices of consumer goods are rising and there is a backlash from the consumers– the decline in the sales of detergents, nappies and luxury cosmetics.

Starbucks is seeking to turn around its fortunes after alienating customers with price increases. Even the stores are not posh. And there are lengthy queues. A revival is difficult in inhospitable trading environment.

Perhaps, the Chinese operations could be wound up. And the organisation should remember ‘retail is in detail.’

The staff uniforms have been changed. It is just incremental change. The baristas write notes on the coffee cups. It is not an effective move. There is a long waiting period –the sequencing should be improved. The little things should lead to a better consumer experience.

Americans buy Pepsi as if the country is in recession — they buy expensive large packs at the start of the month and smaller lower-priced packs as the funds deplete towards the end of the month. Thus, consumers are reining in spending.

Starbucks would not like to use technology to make stores more efficient. There will be more baristas in the stores to speed up coffee making. Ther should not be a backlog of online orders.

Cafes should be more inviting. This should be done in a cost-effective way. A late afternoon menu should serve sparkling drinks.

The prices should be maintained. Starbucks is a ‘simple everyday luxury.’

The crux of the problem is that buyers are becoming thriftier. Discerning consumers prefer a quickly made beverage in a tidy environment, with a pleasant word from a barista.