AI Has Already Surpassed Ninety-Nine Percent of Humans: Prof. Deepak Garg on the End of Mediocrity and the Golden Age for Problem Solvers

Prof. Garg opened with a historical anchor. In 2001 and 2002, when he lectured on neural networks at Indian engineering institutions, the dominant view in the field was clear. Computers could only perform tasks that could be mathematically formulated. Creativity was considered an exclusively human domain. Writing poetry, painting, composing music: all of these were assumed to be beyond the reach of any computational system.

We have been proven wrong

Prof. Deepak Garg, Vice Chancellor, SR University Hyderabad

AI is now creating better art, faster music, and more beautiful poems than many human experts.

Prof. Deepak Garg, Vice Chancellor, SR University Hyderabad

The example he gave was concrete. The music his students consume on streaming platforms is no longer necessarily human-composed. Algorithmically generated music has become difficult for trained listeners to distinguish from human composition. The art on the visual platforms students browse, the poems and short fiction they encounter, the writing they consume on social media: in each domain, AI systems are now capable of generating output that competes with or surpasses much human work in the same form.

Prof. Garg observed, with his characteristic dryness, that what remains uniquely human is increasingly limited to the sharing of emotions such as love and empathy. And even there, he noted that people now turn to AI chatbots for relationship advice and parental guidance. The boundary between what humans do and what AI does has compressed faster than anyone in 2001 predicted.

The next section of the keynote was a structural warning. Domestic appliances of an earlier generation, the washing machine and the refrigerator, were built to serve humans. The current generation of intelligent systems has reversed the relationship. Humans now adapt their behaviour to serve the algorithm. Students entering the B.Tech in AI and Machine Learning at Parul University study the algorithmic mechanics that produce this reversal in detail.

We are assisting machines; we are becoming slaves.

Prof. Garg

The examples Prof. Garg gave to the students were immediately recognisable. No one travels anymore without Google Maps. No one watches YouTube without algorithmic video recommendations. The phone gets opened for one specific purpose. By the time the user puts it down, they have been distracted into watching content they did not seek, purchasing products they did not need, and consuming information they did not request.

We open our phones to do one thing, but the AI distracts us, and we end up buying things we did not even want

Prof. Garg

Prof. Garg extended the observation into domains that students might not have considered. Dating and matrimonial platforms use AI to determine who is compatible with whom. The human element of judgement, intuition, and direct evaluation of another person has shrunk. The decision of who to marry, increasingly, is mediated by an algorithmic match score. The pattern is general. Across high-stakes life decisions, AI has become a silent advisor that few users understand.

Also Read: Rethinking digital security with Prof. Ashutosh Dhar Dwivedi

One of the sharpest sections of the keynote concerned what Prof. Garg called the 1 percent rule. AI systems, in his characterisation, are probabilistic rather than deterministic. They produce outputs based on statistical likelihood rather than guaranteed correctness. Even the best contemporary AI systems carry an error rate. The error rate may be 1 percent or even less, but it is never zero.

It is a probabilistic technology and not a deterministic technology.

Prof. Garg

Prof. Garg illustrated the rule with mathematical specificity. If AI is only 1 percent inaccurate, and approximately 100 crore (one billion) Indians use AI systems for medical advice, financial guidance, navigation, and decision support, then 1 crore (ten million) people receive misleading advice from those systems on any given query type.

A small 1 percent mistake in AI can ruin the lives of millions of people by giving them bad medical advice, wrong directions, or false facts.

Prof. Garg

The conclusion Prof. Garg drew was structural. Humans cannot offload final judgement to AI systems. Every AI output requires human verification. The verification is not optional. It is the difference between a useful AI-assisted decision and a catastrophic AI-trusted error. Rickshaw drivers, shopkeepers, students, and doctors all use ChatGPT and similar systems daily. Very few of those users understand the probabilistic nature of the underlying technology.

Also Read: Pragya Labs giving training to students on simulators.

Prof. Garg addressed the audience’s most frequently asked question directly. Will AI take their jobs? His answer was both more nuanced and more honest than the standard public discourse. Simply knowing how to interact with AI, in his framing, does not constitute a job-defensible skill. A primary school student can type a query into ChatGPT and get an answer. The skill of basic AI usage has been democratised to the point of irrelevance for competitive employment.

The mediocre people, their career is already over.

Prof. Garg

The reason is operational. Companies no longer need mediocre workers because AI can produce mediocre output faster and at lower cost than any human. What companies still need, and need urgently, are deep domain experts. Physicists, biologists, mathematicians, mechanical engineers, and other professionals who understand the fundamentals of their fields are now in higher demand than ever before. The reason is the 1 percent rule. When AI generates new code, designs a new drug compound, or launches a missile guidance system, a human expert must verify whether the output is correct. The verification requires deep, foundational knowledge that AI cannot replace

Prof. Garg’s structural observation for the Parul University students was direct. Those who understand the fundamentals of their domain and also know how to use AI tools are now earning multiples of what their peers earn. The combination is what the market values. Pure AI fluency without domain depth is commodity. Pure domain depth without AI fluency is increasingly inadequate. The intersection is where the careers are. The B.Tech in Computer Science Engineering, the B.Tech in AI and Machine Learning, and the M.Tech in Computer Science Engineering at Parul University all build the foundational depth that the intersection requires.

Check out: Faculty of Engineering and Technology at Parul University, programes that prepare you for all the new track of researches.

Despite the warnings, Prof. Garg framed the moment as historically extraordinary. Earlier generations of Indian students depended on family connections, wealthy relatives, and access to the best teaching institutions to succeed. The barriers were structural. Resource constraints determined who could enter which careers. The opposite condition now applies.

Today, a student from a small village only needs a normal laptop and an internet connection to build a company worth billions of dollars.

Prof. Garg

The educational resources that once existed only behind the walls of elite universities are now freely available online. The compute power required to run sophisticated AI systems is accessible through cloud platforms at student budgets. The market for genuinely useful AI-built products is global from day one. A student in a tier-three Indian city with the right combination of domain depth, AI fluency, and problem-solving discipline can build at a scale that was structurally impossible twenty years ago.

Parul University’s incubation infrastructure aligns with this trajectory. PIERC, the Parul Innovation and Entrepreneurship Research Centre, has incubated 254 start-ups, extended over ₹20 crore in funding, and supported over ₹40 crore in revenue generation across the portfolio. Students with the combination of skills Prof. Garg described have direct institutional infrastructure available.

The most engaged portion of the keynote was the open question and answer with the student and faculty audience. Five substantive questions were raised, each addressing a different aspect of the AI transition.

In response to a question from physics faculty member Sonal Pujara, Prof. Garg cited AlphaFold, the AI system from Google DeepMind that solves the protein folding problem. Where a doctoral student previously spent five years studying the structure of a single protein, AlphaFold now characterises millions of proteins in seconds. Prof. Garg extended the example. Complex mathematical and physics equations that defeated human researchers for decades are now being solved by AI systems. The recommendation he gave researchers was to use AI to read scientific literature and draft research papers, freeing time for the experimental and analytical work that AI cannot yet replace.

A student running his own AI start-up asked Prof. Garg about the next wave of AI tools. The answer named the shift directly. Conventional AI tools including ChatGPT respond to queries with information. They suggest, advise, and provide content. The emerging category is agentic AI: AI systems that act as autonomous agents on the user’s behalf. Prof. Garg used a Singapore travel example. Conventional AI may suggest things to do in Singapore. Agentic AI will monitor weather forecasts, compare airline ticket prices, book the flight, pay from the user’s bank account, reserve the hotel, and adjust the itinerary in real time as conditions change. The decision is delegated. The execution is automated. Students building businesses in 2026 need to be designing for this shift.

Another student asked about layoffs at major technology companies and the future of research careers. Prof. Garg did not soften the reality. Companies including Meta and Google have laid off thousands of engineers. The reason is operational. Code that previously required ten programmers can now be produced by two humans with one AI assistant. Prof. Garg gave a particularly striking statistic. Humans took thirty years to build a computer language compiler. AI recreated equivalent compiler code in twelve days. Despite the layoffs in routine coding, research is expanding rather than contracting. AI now detects cancerous tumours three years earlier than human radiologists. AI tests thousands of drug compounds in seconds, allowing pharmacologists to focus their physical testing on the most promising two or three candidates.

A particularly inquisitive student asked Prof. Garg what would come after ChatGPT. The answer named two technologies. The first is quantum computing. Conventional computers process operations sequentially. A single quantum computer, once it reaches operational scale, will be equivalent to all the computing capacity present on Earth today. Calculations that take classical computers years will resolve in instants. The second is bio-engineering and DNA editing. CRISPR technology allows the cutting and modification of human DNA sequences. The technology can, in principle, eliminate hereditary diseases before a child is born. The combination of these two technologies will reshape what is possible in computing and in biology simultaneously.

The final question came from a student thinking about human history at the longest timescale. Humanity survived as a species for millions of years through creative imagination, the student observed. He asked whether AI would eventually surpass human creativity itself. Prof. Garg answered with honesty.

It has already surpassed 99 percent of humans who are not, whose IQ is lower than a certain level.

Prof. Deepak Garg, in response to the final student question

The audience absorbed the answer in silence. Prof. Garg’s closing observation was that humanity is already addicted to its mobile devices. The boundary between human creativity and AI creativity, in the framing he offered, is not a stable wall. It is a moving line that AI is crossing in different directions at different speeds.

Prof. Garg closed the session with a direct message to the first-year B.Tech students in the room. The world is moving too fast for any university to keep pace with through curriculum updates alone. The students cannot rely on their teachers or their institutions to prepare them for what is coming. They have to prepare themselves. The generation before them, in his framing, can guide and mentor but cannot complete the work.

You do not become mediocre only because your teachers are bad or maybe your university is whatever.

Prof. Deepak Garg, closing remarks

The take-home message of the keynote was a structural one. AI will control much of the world’s routine work. The humans who will control AI in turn are those who know the fundamentals of their domains, reject mediocrity, and solve a new problem every day. Parul University students in the room left with a clearer picture of what their decade looks like and what the first principle of survival within it actually is.