How should humans respond to advancing artificial intelligence?

Ever since it sprung from tech-intelligent fora to our mobile phones in the form of ChatGPT, AI is suddenly as ‘real’ as it can get. Within months of its launch, besides its mass adoption, reams of digital paper have also been spent documenting its superhuman uses, promoting it as a quick-fix panacea to all problems involving a faculty that computers had not breached till just now—Critical Thinking. Technology, to date, could digitize and reproduce innumerable books. Still, the ability to skim through thousands of written words to produce an original formulation, until recently, was under the sole purview of humans. That barrier has been breached. Now there are enough screenshots and videos to show the magical tricks it can pull out of a hat, from market analysis, decryption of ‘Fedspeak’ (on.ft.com/3Hryx60) and sentiment analysis to book summaries, financial planning, website building, economics research, etc. It thinks, plans, decides outcomes, creates sub-tasks and auto-downloads resources to fulfil them.

Amidst this flurry, let us not forget that AI is human-ordained. Creativity, in its truest essence, was divinely bestowed on us. What AI today does, at best, it does by leveraging the output of human creativity over thousands of years. Words, music and art are all made by humans and learnt by AI. Now, as we stand at a juncture where the winning entry in the Sony world photography contest (bit.ly/3VzCk7f), unknown to its judges, was AI-generated, we are looking at a different creative future—one that’s hard to predict.

Man is surely a survivor. AI is surely a gargantuan computing force. Neither is an immaculate predictor of the future. More so when, time and again, an emphasis on economic consequences precludes a discussion around social ones. The economic fears about job losses are not misplaced. As someone recently told us, “ChatGPT is like a smart intern." Basic copy editing, content creation, websites and research tasks can be outsourced to AI even today, let alone after humanity has put its heads together to make it reach the stars. Musk already says that a ‘universal basic income’ (bit.ly/3LHnS9B) will have to become a reality as some jobs turn obsolete. In a consumerist pro-profit market, with requirements of short timelines and a large quantity of work, AI is like manna for its users. This parallels the human story, as machines first replaced manual work—ending many jobs, but creating many more.

But, this time, the survival instinct of man will be tested against the self-spawning and ‘thinking’ power of machines. Earlier, technology did our bidding. This time, we are not sure.

The use of machines freed humans to pursue more ‘thoughtful’ ideas. It seemed sweet, but look at us today. We generate more stuff than we can consume—mountains of content, for example, of poor quality. Already, the attention span of a young adult is a fleeting moment. We struggle to find that one rare virtue that will soon be in demand once our AI lords start producing drivel—discernment.

Unlike the tech of yesteryears, AI has a quality similar to bio-particles, a la the covid virus. It can decide for itself and self-propagate, which can lead to catastrophic consequences in the absence of discernment. Meanwhile, free from the necessity to think even basic things through, humans will have a lot of time. But is that good for humanity? Most kids today find it difficult to produce a beautifully written or thoughtfully analysed work, let alone written by hand. With AI at their disposal, mandated by the labour productivity generation required by corporates, their learning curves can be packed in suitcases and put away.

All good things take patience and focus, like the works of Tolstoy, Valmiki, Shakespeare. Or paintings by Rembrandt and calculus by Newton. It is this base that AI is learning from today, but 50 years down the line, in the absence of a human creativity-facilitative ecosystem, will AI learning from AI be good for us?

Humans and accountability have had an interesting relationship. If ‘distance from direct action’ is mapped on the X-axis of a graph and ‘accountability’ on Y, the farther humans are on the X count, the less they feel directly accountable. This is seen in the Nuremberg trials for war crimes, or the Trolley problem (bit.ly/40R0ECq). We already know this through our everyday experience: think of a signature in ink versus a digi-sign versus checking an ‘I agree’ box. With further action-distancing as AI takes decisions, we remove ourselves from the ethical compunctions we would have if we were the deciders. Say, for an “improve firm profitability" instruction, AI sub-tasks the termination of 50 employees. Advanced intelligence without discernment is a potentially lethal weapon, even in the human case, let alone if the AI is opaque (on.ft.com/3AEadKm) and data-corruptible.

We asked ChatGPT what it thinks about the disadvantages of AI. The top answer was “lack of creativity and intuition", followed by “ethical concerns". That should be a signal. AI is a great tool so long as humans with a conscience are its master.

V. Anantha Nageswaran & Aparajita Tripathi are, respectively, chief economic adviser to the Government of India and director with KPMG.These are the authors’ personal views.