Blog

Autonomous cars tried to be as safe as aeroplanes. However, this has been belied in 2024 — a harmless plastic bag caused traffic disruption by a Chinese robotaxi. Apple has almost given up its plan to build self-driving electric car. Hyuandai and Aptiv halted their joint venture in this area. GM too abandoned Cruise self-driving car project. Elon Musk unveiled promised Tesla robotaxi but it is doubtful whether it will be successful on the roads.

In public mind too, the esteem commanded by autonomous cars have come down. It is treated as futuristic daydreaming. Some products attract mockery and even hate. Google’s parent company Alphabet recalled some 672 vehicles in May. They required a software update, as one Waymo car had hit a wooden pole in Phoenix.

As compared to autonomous cars, the record of aviation in terms of safety is enviable. There is a safety culture within aviation. A plane is certified as airworthy after meticulous tests over a period of time. Once it starts flying, it is subjected to maintenance routine carefully designed. There are checks for the smallest defects and these are carried out by experienced engineers. The checks are extended to the entire fleet, and planes are grounded for this purpose. There are crashes, but they are studied over years to get clues about what went wrong, so that similar things do not repeat in future. Of course, Alphabet’s recall of 672 Waymo vehicles for software update is akin to treating the fleet with the airlines like care and setting aircraft manufacturers’ standards.

Another lesson auto industry should learn from aviation is the sharing of data — safety advances much faster if the data is shared among competitors. Auto industry treats such data as proprietary, and would use it to tread its own path. Tesla has tied up with Baidu in China to follow an aviation-like model, since there is already Apollo robotaxi business of the Chinese company.

It is best to learn from the aviation industry. Driving should cross the trials in a few cities and spread to a vast area in future. Silicon Valley is known to move fast and break things. However, here what is being broken are the human bodies and hence one has to move at a sedate pace.

We are aware of the strike of the writers and actors of Hollywood in 2023, which brought film making to a standstill. The issue was the anxiety caused by the advent of generative AI which made inroads into the writing and editing, and the possibility of it making further inroads into other areas of film making. There is always an inherent resistance to change. Over a century ago, Charlie Chaplin dismissed the emergence of talkie, saying it ‘I give it six months.’ Still, it is forgotten that cinema is a malleable medium. It adapts to the new technology. The new AI technology juggernaut in entertainment industry cannot be stopped. Facebook has rolled out Movie Gen.

AI lowers the barriers of entry into the industry allowing greater diversity of voices and perspectives in film making. AI could make it possible the customization of content, thus opening a new revenue stream. Film makers can tie up with AI companies to train a new model, e.g. Lionsgate partnership with an AI company.

The tie up with AI presents a rosy picture. However, there are issues regarding job replacements. AI should be compatible with existing jobs. Will it snatch control from the hands of the creatives and studio owners? Will AI thrive at the cost of originality? What will be the impact of AI on environment?

Since Charlie Chaplin’s misgiving about talkie, a hundred years have elapsed, and another revolution looms over the industry. It is better to put guardrails to counter its ill effects, rather than putting on blinkers about its existence.

We welcome EVs as game changers in our fight against climate change. There is a claim that EVs reduce carbon emissions and promote sustainable transportation. Government promotes such transportation.

Whether EVs really reduce carbon emissions require a closer scrutiny Apparently, they seem to be environment saviour. However, this ignores its production process which require manufacturing of lithium-ion batteries.

EV batteries have two main components — lithium and cobalt. The extraction of these elements has deep environmental and social costs. Lithium is sourced to the extent of 70 per cent of the world’s requirement from the reserves concentrated in the Lithium triangle of Argentina, Chile and Bolivia. The mining operations are water intensive, and these are arid regions. Cobalt is mainly sourced from Congo. There are hazardous working conditions, child labour and corruption in cobalt mining. There are geopolitical tensions too. The demand for these ingredients is rising continuously. It creates scarcity and there are ‘resource wars’ amongst nations since those nations which control fuel gain power.

The production process of EVs generate 60 per cent more carbon emissions than ICE vehicles, mainly due to battery production. We are shifting emissions from the tailpipe of vehicles, to the factory floor.

Another issue is the electricity used to charge these batteries. In India, 70 per cent electricity is thermal electricity. Thus, EV that claims zero emission has a carbon-intensive lifecycle. EVs charged by thermal electricity can have life cycle emission marginally lower than modern ICE vehicles.

Though India focuses on renewable energy sources, it constitutes only a fraction of total energy generated. Till we decarbonize the entire energy grid, the environmental advantage of EVs remains questionable. Instead, charging EVs through electricity generated by fossil fuels undermines the efforts to counter climate change.

An EV’s battery lasts from 8-15 years and its disposal is another environmental challenge. E-waste generated poses a risk of contamination. Improper disposal adds to the accumulation of toxic substances such as lithium, cobalt and nickel.

EV infrastructure of charging points in India must have at least 10 million charging stations by 2030. At present, India has only 5000 such stations. Another issue the proper upkeep of the charging network. The costs of maintenance may be passed on to consumers. It adds to the ownership costs of EVs.

Instead of private ownership of EVs, India could think of having electrified public transportation. We can also explore other fuels such hydrogen for clean mobility. Even solid-state batteries could be considered.

The global auto industry is in turmoil. The whole automobile industry has been hit hard by two meteorites — Chinese automakers and electrification. Some automakers are abruptly departing from markets and some are closing down their plants. Some continue to operate but at less than their capacity. Many makers write down their investments and sell off their stake in projects.

China has built amazing auto manufacturing capacity. It can manufacture more than 50 million passenger vehicles of all types. It has also emerged as a major hub for making EVs. It accounts for two-thirds of worldwide sales in 2024 and more than 90 per cent of its growth. Its domestic sales of EVs have reached 11.2 million. It can produce only half of this number there. China also dominates the underlying supply chain.

The traditional auto industry is challenged by electrification. They are facing competition in China. And outside China, their business is affected by the Chinese exports. Chinese manufacturers and suppliers dominate the EVs.

Other countries cannot turn a blind eye to the Chinese invasion of the market owing to its massive capacity. The US has already imposed high tariffs and entry barriers on the Chinese products. It is a protectionist policy of the US.

The US is a large market for automobiles. The US market demand is for pickup trucks and SUVs. The US companies forays into EVs and autonomous vehicles is slow. The US companies focus on fuel economy. But the US market is mature market. It is high margin market. Vehicle ownership costs including funding and insurance have reached a natural limit.

Electrification has also changed the vehicle architecture. Branding adds value, but most of the value resides in engine. EVs commoditize battery and electric motors.

The changes in the automobile industry force them to merge together to consolidate their position.

The world witnessed teleportation for the first time in 1993 when IBM put forward quantum state transfers in, unlocking the ways for future breakthroughs. In 1995, Caltech and Wales University researchers teleported photos through co-axial cables. It was a major breakthrough. There was an enigmatic link between the particles, which Einstein called spooky action at a distance. This is the foundation of teleportation. It is quantum entanglement.

In 2002, researchers at the Innsbruck, Austria teleported particles without any physical connection. This is a crucial milestone. In 2017, Chinese researchers teleported photos from earth to a satellite orbiting 186 miles above. It was an achievement. In 2012, Austrian researchers skipped cables while teleporting photos through open air between two landmasses (Canary Islands).

In 2019, Google showed that quantum circuits are superior to supercomputers. They solve a problem 10,000-year-old in just 200 seconds.

Scientists have yet to figure the perfect medium to transmit quantum states with fidelity and efficiency. Should the medium be light or radio waves or space?

Human teleportation is still a dream. Will this destroy the original self? There are issues of identity and ethics.

Alzheimer is a progressive brain disorder linked to the changes in the brain, including the buildup of amyloid beta plaques. Amyloid is natural protein made by the body. Its excessive buildup can form plaques in the brain. It is believed to contribute to memory loss and other cognitive issues.

A monoclonal antibody Lecanemab (lequembi) has been approved by the FDA in early Alzheimer treatment in 2023. It has been developed by Eisai and Bigen. In July 2024, FDA approved Lilly’s Kiunla (donanemab-azbt). These drugs clear the excessive buildups of amyloid plaques. The treatment is approved for adults in early stages of Alzheimer.

In India, there are issues of access and affordability. The performance of drugs will be watched over the next few years.

A previously approved drug in 2021Aducanumab has not shown promising results in the reduction of beta amyloid plaques in brain scans. There was no improvement in patients’ condition.

Prior to transformer, language processing was done by recurrent neural networks (RNNS) and their refinements long-short-term-memory networks (LSTMs) and gated-recurrent-units (GRUs). These processed text word-by-word or token by token. These passed along a hidden state that encoded everything read. It was an intuitive process.

There were shortcomings in these architectures. They struggled with very long sentences. The context at the beginning faded at the end of the paragraph. It was difficult to parallelize, since each step depended on the previous one. Without being stuck in a linear rut, the field needed a way to process sequences.

Good Brain researchers at Google wanted to change this dynamic. The simplest solution was to forgo recurrence. The model, instead, should focus on every word in a sequence simultaneously and to figure out how these words are related to each other.

Attention mechanism enables us to focus on the most relevant parts of the sentence, leaving the baggage of recurrence. The result was arrival of the transformer — fast, parallelizable and good at handling context over long stretches of text.

What was the breakthrough idea? Attention. Transformer elevated attention to the position of a hero of the show.

There were brainstorming sessions. Attention mechanism made translations more effective. The architecture of transformer has two main parts — encoder which processes input data and creates meaningful representation of that data, using self-attention and simple neural networks and decoder which focuses on previously generated output, using information from the encoder.

What is so brilliant in this design? It enables training on huge datasets quickly and efficiently.

The architecture was publicly released and became an experimental tool all over the world. This led to the realization of how powerful the transformer is. The idea that transformer has revolutionized the whole field of AI slowly sunk in.

Google published 2017 paper — Attention Is All That You Need. It outperformed all previous models. It led to a wave of innovations. Google itself designed BERT — bi-directional encoder representation from transformers. NLP became much more efficient. OpenAI took the transformer blueprint and made GPT — generative pretrained transformers. They had GPT 1 and GPT 2. The models produced human-like text. In late 2020, ChatGPT was introduced based on GPT-3.5. It generated coherent text, could translate languages, could write code and even produce poetry. Machine’s ability became a tangible reality. AI-assisted creativity became an everyday thing.

The transformer is not confined to text. Attention mechanism could be extended to different types of data — images, music, code.

Software frameworks (TensorFlow and PyTorch) incorporated transformer-friendly building blocks. Models were scaled up by increasing the number of parameters in a transformer and the size of its training dataset. There was a race for bigger models, more data, more GPUs. The field attracted massive investments.

Training such models consumes compute time and electricity.

There are issues of synthetic data, copyright, misinformation, deepfakes and ethical deployment. The governments propose to regulate the field without stifling innovation.

The Attention Is All That You Need paper is a testament to what open research can do for the world. The paper allowed every researcher to build on its ideas. It is the spirit of openness. Transformers are the default backbones of NLP. Should we go beyond attention to find new models? The field is not stagnant.

Technology companies are optimistic about great leaps, and more revenues. Yet, meaningful revenues have to wait till 2026.

In 2024, there was a sudden fall in Nvidia shares due to a glich in its new chip. It was resolved and yet investors did not remain assured. They soon recovered sobriety. The incident demonstrated that deep within there is lurking anxiety about AI. Anytime something goes wrong, there is a knee jerk reaction.

It does not augur well for 2025. AI development has slowed down, or possibly it is crawling. AI champions are selective about the words they choose. Sundar Pichai, Google CEO, feels that the ‘low-hung fruit’ has already been picked while going to the next level we require more insightful breakthroughs.

Sam Altman was optimistic about OpenAI reaching AGI. Yet he admitted that it would not matter much as some previously thought. Of course, superintelligence will be a disruptor, but it is further away. Sora, a video generator software, is too expensive. It has some reasoning faculty but mostly it is all bells and whistles. It is an iterative model. ChatGPT had the wow factor when introduced 2 years back, but it has faded away. Apple Intelligence has not impressed the users, and the summaries produced by it are severely mediocre. Apple would have liked to be not the first but the best, but that has not happened. Anyway, AI would not become Apple’s USP.

As we move into 2025, LLMs are being loaded with more and more data, but unfortunately, the limit has been reached, and the models are not being proportionately better.

AI and generative AI were overhyped, and investors are worried about the prospects. The jury is very much out. There is a search for lucrative use cases. However, we are not advising the selling of AI stocks, You should hold them. A year may not be pan out as you wish. Maybe, you will wait for 2026. Wall Street should be prepared for AI winter.

Futuristic technology feels like science fiction. Google has unveiled, as we have observed in a separate blog, its quantum computer chip called Willow. Google’s CEO Sunder Pichai has floated an idea of taking quantum computing to space.

Traditional computers are very slow as against quantum computers. There are several challenges in quantum computing. Quantum chips are isolated so that there is no electromagnetic interference. Besides, quantum chips must operate at extremely low temperatures. These conditions are difficult to satisfy on the earth. This hinders their scaling.

Elon Musk’s company SpaceX has introduced Starship, and he is ambitious about the lunar and Martial landings, and colonization in future.

Sundar Pichai suggests that quantum computers must be sent into space to take advantage of conducive conditions. There is no interference of the earth in space. The stable low temperature also could be maintained. Thus, if the quantum systems are in orbit they can process data and conduct research. Elon Musk has responded positively to this proposal.

The conversation was not restricted to quantum computing but was extended to solar energy. Pichai hinted at future technologies such as Dyson spheres.

The ideas are still nascent. However, they have potential benefits — climate modelling, healthcare and material science.

Entanglement is a crazy phenomenon. Apparently, it allows particles to communicate over vast distances, faster than the speed of light.

Though there is connectivity between entangled parties, these do not necessarily share information between them.

In quantum mechanics, a particle is not really a particle but a cloud of fuzzy probabilities. These probabilities describe where this particle could have been placed, when we look for it.

These fuzzy probabilities are in fact quantum states. If the certain circumstances occur, we can connect two parties in a quantum way. Therefore, a single equation describes both sets of probabilities at one and the same time. This by itself is called entanglement.

Quantum state particles show similar properties. We know electrons, the sub-atomic particles, revolve around the nucleus of the atom. Let us look at their quantum spin. The spin could be in one of the two states — either up or down. If these two electrons are entangled, their spins show correlation. The entanglement can be planned in a certain way, so that the spins are always opposites of each other.

Examine one particle randomly. There is random spin that points upwards. Can we predict about the second particle? The quantum state has been carefully arranged. Therefore, we are sure that the second particle must be pointing downwards. The quantum state was entangled with the first particle, as soon as one declaration is made. That means both the declarations are made.

What happens if the second particle is at a distance? In quantum theory, as soon as one particle is chosen, the partner particle instantly knows what spin to be in. Apparently, it is communication faster than light.

If two different persons take measurements for particles, the person measuring particle A knows the spin of particle A. Or else the other person tells you about the measurement. In both the cases, there is no transmission. Either you measure or wait for the other person’s signal.

In case of two connected particles, there is no advance knowledge of anything. The person knows how the particle is behaving but I communicate it slower than the speed of light.

Thus, entanglement process, though instantaneous, the revelation of it is not instantaneous. The older communication method prevails to piece together the correlation of the quantum entanglement.