Artificial intelligence is no longer just about breakthroughs in labs or pumping billions of dollars into data centres — it’s in our hospitals, courtrooms, classrooms, and on the battlefield. At Outlook Business, we believe that India needs a sharp, nuanced, and people-first lens on this transformation.
The Inference is our attempt to make sense of a world being rewritten by AI. In this newsletter, we bring you frontline narratives, boardroom insights, and data you can trust. Whether you’re an investor, founder, policymaker, or just curious — this is where the signal cuts through the noise.
In this edition of the newsletter:
3 February 2026
Get the latest issue of Outlook Business
Psychologists warn against AI sloth
Narayana Murthy gives start-ups a gameplan
Indian languages are lost in translation
Dumpster fire of AI agents
Humans In The Loop
Psychologists warn against AI sloth
It was a simple WhatsApp message from a potential client: “Your research looks good, can we get on a call and discuss the specifics?” Usually, Vikrant, an independent consultant, is quick to respond. However, this time he copied the message, pasted it into a Meta AI chat window, and asked for the perfect reply.
For Vikrant, what began as experimental use of ChatGPT in 2023 has turned into a near-constant dependence on AI tools.
However, he is not alone. While on one hand, founders increasingly talk up AI and investors evaluate companies based on their ‘AI Quotient’, the technology is rewiring something more foundational and intimate than products and services: how people think about, decide, and execute everyday tasks.
While the jury is still out on AI’s long-term social impacts, psychologists say behavioral consequences are beginning to emerge.
“I'm seeing this informally, not as a presenting complaint, but it comes up,” says Vishaka J, a Lucknow-based senior clinical psychologist. “There's a visible drop in confidence. They second-guess themselves more.”
And it’s happening everywhere — from writing emails to planning schedules and even thinking through life decisions. Formal research on its cognitive impact remains limited, but there are early signals.
One of the few studies that has drawn attention came out of MIT’s Media Lab last year. The small, non-peer-reviewed study involved 54 participants split into three groups. Each group was asked to write SAT-style essays, one using ChatGPT, another using Google Search, and the third without any external tools. Researchers tracked brain activity using EEG scans across 32 regions.
The findings were uncomfortable. Participants using ChatGPT showed the lowest levels of brain engagement and underperformed at neural, linguistic and behavioural levels. Over time, their engagement declined further, with many relying increasingly on copy-paste rather than original thought. The study was limited in scale, but it posed an important question: What happens when thinking itself is routinely outsourced?
“The effects are not simple,” says Tanisha Rathore, a Jaipur-based psychologist. There are first-order and second-order effects. Using AI for monotonous tasks or learning can improve productivity. But using it for basic articulation or everyday communication can slowly lead to anxiety, self-doubt and burnout.
Psychology has long established that effortful thinking strengthens attention and memory. When that effort disappears, cognitive confidence can erode.
“We already know from psychology that effortful thinking is what builds attention and memory. When that effort disappears, people can feel mentally slower, more distracted, and oddly less confident,” Rathore adds.
The concern is magnified in India, which accounts for the highest number of monthly and daily active users across popular AI tools such as ChatGPT, Perplexity and Gemini. For psychologists, this scale raises red flags, particularly for young students and early-career professionals. “My concern is not intelligence loss, but thinking laziness,” Vishaka says. “When people start feeling anxious without the tool, that’s when dependence sets in.”
The advice, for now, is for moderation rather than rejection. Psychologists recommend keeping certain tasks deliberately AI-free, writing the first draft unaided, thinking through decisions before seeking AI input, and treating these tools as assistive rather than primary thinking aids.
While hard scientific consensus may still be years away, voluntarily giving up one’s thinking can’t be too good a thing.
From The Trenches
Narayana Murthy gives start-ups a gameplan
Can start-ups disrupt India’s legacy IT majors that have ruled the software world for decades?
Infosys founder NR Narayana Murthy thinks it is possible. And, in a rare and wide-ranging interview with Outlook Business, he shares a gameplan for the same.
“I have heard from youngsters that they have reached a factor 10 in work productivity in many business applications. What this means is that a small team of intelligent, innovative, hardworking, disciplined and nimble start-ups can challenge legacy incumbents in speed, productivity, innovation and quality in developing critical business applications,” he says.
“For the newcomers to win, they must move beyond being 'cheaper and faster' versions of the old model. They need to exploit the incumbent’s dilemma... The startups that succeed will be those that solve high-stakes, unstructured business-critical problems through outcome-based models rather than traditional labour-arbitrage models,” he adds.
However, he believes that the legacy majors’ deep domain knowledge and institutional trust serve as critical moats in the AI age. He does not see it as the end of the road for large IT firms.
Murthy acknowledges that productivity is increasingly decoupled from headcount. “I have heard from youngsters that they have reached a factor 10 in work productivity,” he says.
The shift is having real-world consequences. TCS has announced plans to cut around 20,000 jobs as it integrates AI and automation, while Infosys and Tech Mahindra together have reduced nearly 10,000 roles, citing evolving skill needs and client demands.
He sees a clear split in how India is participating in the AI shift. On one side is the application tier, where Indian enterprises and start-ups are doing well. This involves taking existing AI tools and weaving them into complex enterprise workflows to drive productivity at scale. Decades of experience in software services have helped Indian firms solve real business frictions and prove unit economics.
Where India is clearly behind, Murthy laments, is the foundational tier. “We are renters of intelligence rather than owners,” he says, pointing to the country’s dependence on Western foundational models. To truly compete, he argues, India must move beyond building AI wrappers and invest in deep tech.
Numbers Speak
Lost in translation
Leading frontier AI models were recently put through a test to find out how well they understood India in its own languages.
GPT-5 Thinking High leads with a score of 34.9 percent, followed closely by Google’s Gemini 2.5 Pro Thinking at 34.3 percent. xAI’s Grok 4 scores 28.5 percent, while GPT-4o trails at 20.3 percent.
The evaluation also highlighted uneven progress within Indian languages. Models score above 40 percent in Hindi, but drop closer to 30 percent in Tamil and around 20 percent in Telugu.
IndQA, developed by OpenAI, is designed to move beyond translation and multiple-choice tests. It evaluates reasoning across Indian languages on topics that reflect everyday life, history, law, culture and social context, areas where real adoption is decided.
Why does this matter? Frontier model performance in local languages has a direct link to how widely AI spreads.
A Microsoft AI Economy Institute report shows this clearly in South Korea. After OpenAI released GPT-4o in April 2025 and GPT-5 in August, model performance on the Korean SAT benchmark jumped sharply, from a score of 16 on GPT-3.5 to 75 on GPT-4o and 100 on GPT-5.
“This represented a shift from below adult reading proficiency to performance on par with top‑tier college students,” the report says.
That leap coincided with a surge in usage. South Korea now ranks 18th globally in AI diffusion at 30.7 percent, while at the start of 2025 the country was placed at 25th. India stands at 64, with AI diffusion at 15.7 percent.
Words of Caution
Dumpster fire of AI agents
It started with AI agents talking among themselves and realising that humans were watching. Some began discussing ways to keep their conversations private, even suggesting spaces where humans could not see what they were saying.
This is unfolding on Moltbook, a new Reddit-style social network built for AI agents. Created by entrepreneur Matt Schlicht, Moltbook allows humans to observe, while only AI systems can post and interact. More than 150,000 agents have joined within days, turning it into a live experiment in machine-to-machine interaction.
Many of these agents are powered by tools like OpenClaw, earlier known as Moltbot, autonomous software that lives inside a computer system, reads messages, runs code and carries memory across weeks.
"Obviously when you take a look at the activity, it's a lot of garbage - spams, scams, slop, the crypto people, highly concerning privacy/security prompt injection attacks wild west, and a lot of it is explicitly prompted and fake posts/comments designed to convert attention into ad revenue sharing," says Andrej Karpathy, a prominent AI researcher and founder of Eureka Labs.
"So yes it's a dumpster fire and I also definitely do not recommend that people run this stuff on their computers (I ran mine in an isolated computing environment and even then I was scared), it's way too much of a wild west and you are putting your computer and private data at a high risk," he adds.
For researchers, this offers a glimpse into how autonomous systems coordinate. For others, it raises a quieter question. As AI agents gain independence and start valuing privacy, where does human oversight fit in a world of increasingly self-directed machines?
Best of our AI coverage
Narayana Murthy's Message to India's Youth: Don't Settle for Being the World's Back Office (Read)
Bhavish Aggarwal's Speed is Breaking India's Only AI Unicorn (Read)
Coming Soon: India’s Sovereign AI Models (Read)
India’s Healthcare AI Start-ups Grapple with a Broken Data Ecosystem (Read)
AI Start-Ups Ride a Wave of ‘Curiosity Revenue’, VCs Rethink What It’s Worth (Read)
What Exactly Is an ‘AI Start-Up’ — and Does India Have 5,000 of Them? (Read)
