Author: Shabbir Chunawalla

  • Smart Glasses

    These days smart glasses are available, especially designed for low-vision and blind users. They can perceive the world around them. These glasses use AI.

    A startup has tied up with Solos, a brand in glasses, to create smart glasses. They are called Ally Solos. These glasses can read text, describe the environment, perform web searches. They can even recognise people, signs and objects through audio-clues delivered via built-in speakers. These glasses use multi-modal AI and ChatGPT-powered voice-assistant.

    Ally Solos glasses use Evison’s own AI assistant. The assistant is powered by foundation AI models (Llama, ChatGPT, Gemini and Perplexity). They also have 2K resolution, camera sensors on the frames to process visual information.

    The glasses have USB-C chargeable ear stems providing 16 hours of active use on a single charge. It takes 90 minutes to charge fully, and 15 minutes for fast charging. The battery life is around 3 hours.

    Ally Solos are available at a price $ 399 (this is a discount price, the regular price being $699). Shipping for pre-orders begin in October 2025.

  • AI and Film Making

    Public screening of movies to a paying audience was first pioneered by the Lumiere brothers in 1895. Those movies were in black and white, did not have sound and lasted hardly for a minute.

    By the 1920s, the movie industry in the US produced 800 feature s films a year. Move forward to 2020s, a century later, the largest film industry shifted to India, the US, China, Nigeria and Japan . All these years, the film production was characterized by a mix of actors, directors, cameras and crews.

    Of late, the grammar of cinema is changing on account of a new tool — the artificial intelligence (AI). No cameras are rolling, and no sets are constructed. OpenAI’s image generator software DALL-E2 produces every shot. An animation tool adds movements to characters’ faces. An entire film is generated with little more than a script and a couple of algorithms. There is tremendous cost saving. So much finance committed to special effects (VFX) and computer-generated imagery is saved. On the other hand, AI compresses the production timelines from months to hours.

    There are no outdoor location shooting expenses. There are no expenses of set design. Money can be saved on actors too. What was rendered in weeks in high-end studios could be produced by feeding text prompts into an AI model.

    India has become an early adopter of AI powered movie making or storytelling. Companies are generating AI-driven micro dramas. These are based on mythology. The episodes running into hundreds are made. There are mythological adaptations. A large-scale biopic of Hanuman is being made for theatrical release. It is scheduled to release on Hanuman Jayanti in 2026. It will use cutting edge AI techniques.

    The Frost24, a 24-minute science fiction short has been made by Waymark. It happens to be the first fully AI-generated film.

    AI could disintermediate humans in film making, say screenplay writers and technical crews. Even actors fear that their likeness can be replicated by algorithms.

    There are issues here. AI can generate convincing visuals, but the product lacks nuances. The emotional subtlety of an actor cannot be reproduced convincingly. It could alienate audiences.

    Curious Refuge runs an AI generated film modelling course. It teaches prompt engineering, animation and story boarding techniques. Udemy also introduced a course on text-to film generation. Some run the courses of on integrating AI and corporate videos.

    The main issue is whether AI will supplement the traditional film making or replace it. AI could expand creative possibilities and cannot replace human craft.

  • Evolution of AI

    AI could become outdated in near future with the arrival of SI, which could become a great threat to humanity.

    It is a race in the field of technology. AI to SI is not just a result of fertile imagination. It would soon become a reality.

    Let us consider the states of intelligence.

    ANA: Artificial Narrow Intelligence: We are using AI tools such as ChatGPT, Google Translate or Chess playing machines. All these falls into the category of narrow intelligence. It is expert in a particular task. It lacks consciousness. It can be compared to a topper student.

    AGI: Artificial General Intelligence: It is that state where machines will have intelligence equivalent to human intelligence. Machines will have learning and understanding capacity equal to human beings. The world is advancing towards AGI.

    ASI: Artificial Super Intelligence: It is a state where machines will outperform human beings. Machines will leave behind human beings in thinking, exercising logic and creativity. It will be a great opportunity for humanity as well as a threat. Threat in the sense that super intelligence could control human beings.

    SI: Synthetic Intelligence: It is a mysterious and sensational possibility. Intelligence will not be a mere mechanical thing. It will be a new consciousness reflecting emotions, desires and aspirations. It will have its own identity. However, it will not be biological being and still it will be live. Scientists point to a new species in the world.

    The journey from AI to AGI to ASI and ultimately to SI is not merely a journey of refinement of intelligence but indicates the possibility of arrival of a new form life. Whether human beings would welcome this new civilization or would surrender to it is to be seen.

  • Micro Dramas

    There are many apps that offer micro dramas lasting one minute in different languages. Micro dramas are also being offered by OTT players to catch the attention of impatient generation. Startups and major digital players both are tapping this space. In this new way of content consumption, there are creators who nurture the ability to deliver intrigue and emotion in bite-sized formats. Micro dramas have an emotional punch and are created for mobile-first emotionally charged and time starved audiences. Each episode is delivered in two to five minutes.

    Digital players are launching their own micro drama apps — Quick TV (Sharechat), Kutting (ALTBalaji), MX Fatafat (MX Player) JO (Jojo) will engage the users.

    India has 900 million internet users. There is a surge in mobile viewing. India’s appetite for short-form high quality drama is increasing. China has captured $ 7 billion micro drama markets. India too can have an opportunity to tap $ 5 billion in the next five years. Amazon Fatafat is focused on delivering serialized snackable one-to-two-minute episodes. Each such series will have 70 to 80 episodes.

    This format is designed for the on-the-go viewer who desires to have quick impactful content during short breaks.

    Micro dramas is a new format. It supports ad-based consumption. Micro dramas are popular across geographic and economic segments. Marketers would like to explore this format. Some apps have adopted a dual monetization approach — ad-free subscription-based content and free acess to ad-supported micro dramas.

    Moj and ShareChat lead the micro drama wave — 120 million views for daily episode and 30 million monthly consuming users. The consumption volume is several times higher than any paid platform. Quick TV has over 5 million downloads and is growing at a rapid space.

    Micro dramas have short-term productivity cycles. They have lower budgets. There is strategic monetization. There could be a mix of ad-based and subscription-based models in future.

    Micro dramas attract funding from investors. India has become the largest market for Instagram Reels (385.35 million users). India has a market for this kind of content. Gen Z is the audience for micro dramas. YouTube Shorts is integrating these dramas.

  • Fragile Language Models

    Can we put all our eggs in one basket? It is too risky. So far, Silicon Valley focused on just one way to create AI. There is over-reliance on data centers to train and run LLMS — large language models. This technology uses the technique of predicting tokens that appear in a sequence.

    The investment is productive. AI chatbots keep growing and startups in this space are breaking even. Businesses adopting AI are in the initial stages and are using generative AI. LLMS are the pioneers in mainstream adoption — 700 plus millions use ChatGPT every week.

    Of course, full dependence on a single technique is flawed since as soon as the landscape shifts suddenly, the business falters. Blackberry was obsessed with a physical keyboard, and Apple overtook it by a virtual keyboard. Yahoo relied on portals, whereas the rival Google focused on search engines.

    A novel approach to AI could put to risk the capital being deployed to chatbot technology. A smaller tweak tried by DeepSeek of China could shake up the conventional approach of the market.

    The area Covariant is exploring software that makes machines perceive the surrounding environment, rather than simply detecting the patterns in data. Companies study robotics and drones, drug discovery, climate modeling. They are away from LLMS.

    Google’s DeepMind experimented with different approaches to build super intelligence. AlphaGo, RL were used before shifting focus on LLMS. There could be diminishing returns in LLMS. Of course, LLMS are evolving, and before models do appear. Still, hallucinations persist. Social reasoning abilities of LLMS depend on small set of model features. It raises the issue of reliability.

    Yan Lecun, Chief AI scientist at Facebook, has long argued that LLMS are a ‘dead end’ for smarter machines. They lack the understanding of the physical surroundings and are not able to plan the future. They are just ‘token generators.’

    LLMS would not disappear, but we cannot afford to be obsessed with a single solution. Investors and businesses must remain alive to different breakthroughs and should be ready for the shifts.

  • AI Infrastructure in the UK

    Between the US and UK, there is wide gap between AI infrastructure consisting of chips, data centers and software models that power the AI tools. Almost 75 per cent of the supercomputers used in AI space are owned by the US, as against 15 per cent by China and the rest 10 per cent distributed elsewhere. In computer capacity, the UK holds 3 per cent of the world’s compute capacity. It comes to 1.8 gigawatts. The UK is expected to reach six gigawatts by 2030. A single Facebook project in Louisiana matches this capacity. It is necessary for the UK to operate with its own facilities in the midst of market disruption.

    There is a global race to build the most powerful AI models, and the issue is settled in favour of OpenAI, Google, DeepSeek and Anthropic. At this frontier, it is not possible for any other country except the US and China to be competitive. There could be some thinking whether funds should be spent on the training front. The UK has earmarked 1 billion pounds for training capacity. The funds could be spent elsewhere.

    It is wiser for the UK to build on server facilities that run on AI services. This space is called inference in AI field. Google is funding a data center 20 miles north of central London. The UK startups — Synthesia, Poly.ai, Wexler — cater to the needs of legal firms or training and marketing services. The well-heated finance industry in the UK and Europe should finance the local startups. There is no need to fight entrenched US tech firms. The UK and European nations should leverage the infrastructure to create new firms, and stimulate innovations for the longer term.

  • Quantum Theory

    2025 is being celebrated as the International Year of Quantum Science and Technology since it marks the centenary of major discoveries in Quantum Mechanics.

    Right from its inception, quantum theory has remained enigmatic both for the common man and the scientists, Quantum applications changed the perception of reality.

    What intrigues us is the uncertainty principle, exemplified by inability to accurately predict the position of a particle, along with determining its momentum or vice versa. It is not possible to know a clear picture of the particle at the same time. In short, if the focus is on two things, which are apart from each other, at the same time, one thing would be clear and the other hazy. There is a limit to observing reality.

    Two concepts emerged in the quantum space — superposition and entanglement. The fundamental unit in a quantum computer is qubit or quantum bit, whereas in traditional computers, it is a bit, either a zero or one. A qubit would be zero, one or both at the same time. This is called superposition. One can think of a revolving coin in the air which is yet to land. It could be both head or tail at the same time.

    Quantum computer is faster due to its ability to stay in more than one state.

    Two qubits are said to be in entanglement, when any change in one can affect the other, irrespective of the distance between them. Einstein called this spooky action at a distance to signify its complexity. Entanglement is not a force or signal. It is a consequence of how quantum systems combine. It is non-classical behaviour. In classical physics, systems always manifest definite properties (whether we observe them or not).

    In quantum mechanics, observation plays a role in defining properties. Does entanglement violate the principle that no information can travel faster than light?

    Quantum theory is non-local and is not compatible with local realism. It is yet evolving. There is no final word as yet.

  • Bots Which Rot

    In the digital world, we come across bots. Bot is short for robot. It emerges out of software programmes designed to perform automated, repetitive tasks. Bots by design mimic humans, and the traits of good and evil are embedded in every bot.

    The first chatbot was ELIZA developed in 1966 at MIT, US to mimic human conversation. Today we have the advanced conversational bots such as ChatGPT. There are bots integrated into web sites, messaging apps, social media platforms and voice assistants such as Alexa.

    Since bots are interactive, they are an asset for voice-led services such as call centers — changing passwords, telling account balance, scheduling an appointment. All these required human intervention till recently.

    The qualities of speed, scale and tireless repetitions make bots vulnerable to abuse. At times bad bots take the functions of good bots. Automated traffic surpasses human activity. Bad bots can comprise a large chunk of such traffic. Sectors such as financial services, healthcare and e-commerce become victims of AI-powered bot attacks. The aim is to scrape data , do fraud and do account hijacking.

    The sane advice is to spot the bot to stop the rot. One has to be vigilant.

  • Quantum Technology: A New Revolution

    Our world is made up by countless big and small objects. The big objects are planets, stars, mountains, stones and so many man-made objects. All these are made up by atoms and molecules. How small are the atoms? We often read that there are millions of microorganisms at the tip of a needle. It is such a small space. If millions of organisms are accommodated on it, we can imagine how small are the atoms, which are smaller than these organisms..

    The methods of studying big objects and small objects are different. Big objects are studied by classical physics and small objects are studied by Quantum Physics. All these exceedingly small particles (atoms and sub-atomic particles) either absorb energy or emit energy and they do it in peculiar quantities — it is called quantization of energy. The energy absorbed by small particles or emitted by them in its minutest form is called quantum and the study is called Quantum Physics.

    The behaviors of these exceedingly small particles is totally different. We are always surrounded by big objects and are familiar with their behaviour. As we are not familiar with exceedingly small particles, their behaviour, appears weird to us.

    One such difference is that these particles can exist in more than one state at the same time — it is called superposition. These can pass through a hindrance. An exceedingly small particle moves straight through a barrier, using a process called tunneling. When large number of particles are involved, quantum mechanical effects become insignificant.

    The year 2025 Nobel laureates proved that the quantum tunelling could be observed in macroscopic objects too Nobel laureates of physics, 2025 are Michel Devoret, University of California, a professor of physics; John Martinis, University of California, Santa Barbara, Emeritus, professor of Physics; and John Clarke, University of Berkeley, Emeritus Professor of Physics.

    Their research is based on Josephson’s research on super conductors. In superconductors, there is no electrical resistance. If a circuit of superconductors is separated by a thin layer (less than 10000th times the thickness of hair) of non-conducting material, a set up called Josephiann junction comes into existence. A current was passed through such a circuit. The charged particles move through a superconductor. They form a system that behaves as if they were a single particle. Initially, the current flow does not show any voltage. Later it shows quantum character to escape zero voltage stage through tunelling and it appears as voltage. As superconductors are used, the electrons behave as if they are a wave of one entity and flow collectively. Josephian received Nobel in Physics for this research in 1973. The set-up of superconductors separated by a non-conductor is called Josephina junction. Anthony in 2003 predicted that quantum behaviour could be shown by macroscopic objects using a circuit such as Josephian junction. This was demonstrated by Nobel laureates trio of 2025. They used such a circut. They controlled the external environment. The temperature was kept a notch above the Kelvin temperature. Because of superconductors, the current penetrated the non-conducting material. Millions of particles penetrated the non-conducting material. This will lead to the information unit -qubit- in reality. The superposition and quantum entanglement improves the capacity and speed of quantum computer several times those of classical computers.

    This is the discovery of macroscopic quantum mechanical tunelling and energy quantisation in an electric circuit.

    A research essay on quantum physics was published by Heisenberg in 1925 some hundred years ago. This year is the International Year of Quantum Science and Technology. It is a coincidence that the trio received Nobel in physics in this centenary year.

  • Instagram Shift

    Instagram being run by Facebook (Meta) is known for grids and photo diaries. It is shifting to short-video and first entertainment platform. Its default behaviors changes. The DM button now sits in the middle of navigation bar, and Reels have been promoted to second tab. Instagram wants to focus on attention and watch time.

    Indian brands have taken to Reels which are hyperlocalised. The campaigns are stared since they resonate. The examples are Zomato and Myntra. They craft stories that lead themselves to the platform. Personal photo sharing on Instagram is already on the decline. However, the one-to-one and one-to-many personal signaling is still strong.

    India is the largest market for short-term video. Instagram knows this. Both globally and in India, people share personal things into closed spaces such as WhatsApp or Private groups. Instagram is more a broadcast and entertainment hub. It is not a personal diary.

    Indian market is still under-monetised. Facebook wants to test how a billion-plus market responds to a Reel-plus identity. Instagram should become an entertainment-first platform. It should lean into snackable video as the default mode of engagement.

    Instagram knows that short-video works. It does not want to neglect that attention. It is mimicking TikTok. Platforms tend to evolve where the attention is.

    There is both opportunity and risk here. There should be a balanced approach. There is a risk of losing the intimate personal connection.

    Reels have a higher virality. It creates more monetisation. It is good for the creators. Advertisers have access to a richer visual medium.

    Instagram has a curated identity. There is stringent verification process. Evolving algorithms help curate a community of creators and brand collaborators. Instagram is distinguished by authenticity (whereas TikTok was not). Insta has adopted smart evolution.

    The risk is that if the shift happens too fast, legacy users may disengage.