The Aakhya Weekly #54 | What AI means for your workplace

Jun 30, 2023

In December 1900, a little-known American engineer, John Watkins, took a crack at predicting what the year 2000 would look like. Despite getting a few things wrong (Extermination of all mosquitoes? Peas as large as beets? No more wild animals?), he was surprisingly prescient. He predicted digital photography and mobile phones, television and battle tanks, aeroplanes and bullet trains… and much more. There have been others like him. Jules Verne predicted the moon landing. Stanley Kubrick predicted the iPad. Mark Twain imagined up the internet.

Of course, not everyone is blessed with such clairvoyance. From square tomatoes to underwater frogmen, many have made embarrassingly poor predictions of what the times to come would look like. Even so, their predictions feel easy to understand. They extrapolate from the technological frontiers of their era, make a few assumptions around how humans would behave in their wake, and voila! They’re in print.

This is a luxury that is no longer available to us, for we are entering the era of artificial intelligence (AI). If there is one thing we understand, it is that this technology is unpredictable; that its limits are, as yet, unclear. Its proponents weave utopias, of cascading golden ages wherever the eye can see. Opponents fear an existential risk akin to that of a global nuclear war - one that leads to the extinction of all humanity.

That is merely the technology. How humans shall interact with it is yet more uncertain. As much as one may fear the omniscient titans of Asimov’s imagination, it is humans, in all their murky, confusing glory, that truly test the limits of our foresight. Twenty years ago, who would have guessed that social media would unleash an epidemic of fake news and poor self-esteem? AI will, in all probability, throw up bigger surprises.

And yet, can we afford not to look into the crystal ball? AI has already entered the workplace. Entire industries now stare at automation. And this, we are told, is merely the beginning. How will it all play out? How can we prepare ourselves for the inevitable?

The answers, alas, are not available to us. What we have, however, are a few questions to sharpen our thinking on what is to come.

The foundational question: What is AI, even?

What do you think of, when you think of “AI”? Perhaps you’re addicted to the current wave of generative AI, that has thrown up everything from Bach-like compositions to a Pixar Harry Potter. Perhaps you’re terrified of TikTok-like algorithms that, in a few short hours, understand you inside-out. Maybe you think of robots in gigafactories that churn out cars within minutes. Or chess bots that beat world champions. Or weird robo-dogs that play football.

How does one even begin to wrap one’s head around technology that does everything?

Well, fundamentally, AI is not a technology. It is, rather, a class of technologies that mimic human abilities. Machine learning systems parse through wide arrays of data to find correlations, and perchance, create something resembling abstract concepts. Natural language processing allows machines to understand written and spoken language. Robotics allows machines to move and interact with the physical world. [Computer vision](https://www.ibm.com/topics/computer-vision#:~:text=Computer vision is a field,recommendations based on that information.) allows them to make sense of visual data. In short, there is no one AI frontier, but several.

The things AI can do depends on how these individual technologies grow and play off each other. Midjourney, for instance, forms “foundational models” through “deep” machine learning, and then puts them together in interesting ways in response to prompts, which are, in turn, decoded through natural language processing. Self-driving vehicles learn how traffic behaves through machine learning, and then use robotics and computer vision to move around. The potential of these applications, as well as their shortcomings, arise from these underlying strands of AI technology.

Unknowns, both known and unknown

Broadly speaking, modern AI can sense the world, “understand” it and interact with humans and physical spaces alike. Most human occupations are but some arrangement of these processes. Is this it, then? Should we panic?

Alas, we cannot say for certain. The trajectory of AI is famously unpredictable. In 1956, for a while, we believed that ten people could crack the puzzle of artificial intelligence over a summer. Other times, however, were marked by such pessimism that the entire field came to grinding halt. In 2017, McKinsey attempted to quantify the economic potential of AI. In 2023, just five years later, it revised its estimates by up to 40%. It is hubris to believe we will not make mistakes again.

The holy grail of AI research is “Artificial General Intelligence”. Currently, most AI performs narrow, highly specific tasks – such as parsing through medical research, or tracking the stock market. Contrast this with our own brains, which can, with a little effort, understand or do almost anything in our physical capability. With AGI, we shall bring such malleability to machines. The more malleable AI is, the more completely it might replace us. Our pathway to AGI is still unclear. Perhaps we will never get there. How long we take to crack this puzzle is our biggest unanswered question surrounding the impact of AI on work. Curiously, with models like ChatGPT, we have come surprisingly close – indeed, it is why they seems so human. How close, however, is anyone’s guess.

There may, however, be other questions that we simply have not thought of asking. So far, we don’t completely understand the human mind. Is there something fundamental about us that AI simply cannot recreate? Despite our best efforts, we do not know. There may also be pragmatic, real-world constraints that we have not anticipated. AGI might turn out to be far too expensive for most tasks. The computational power it takes may be well beyond our capabilities. We may not have the data to train it for most practical purposes.

The destiny of our workplaces is tied to the answers to these questions.

What we should anticipate

AI will not enter every workplace in the same way. Different workplaces will take to it differently, with varying implications for their workers.

In the wake of AI, many jobs will be automated, pushing many out of the economy. McKinsey estimates that anywhere between 400 and 800 million people may be laid of by 2030. In a tougher labour market, even those who keep their jobs may see their salaries depleted – especially if AI takes up a large chunk of their tasks.

In other jobs, where several tasks are harder to enumerate, humans will work alongside AI. Many workers, particularly in online labour and gig platforms, are already supervised through AI. This allows enterprises to operate at far larger scales than ever possible, but also limits worker autonomy. Other workers shall use AI to augment their work, increasingly their productivity significantly. AI-enabled enterprise software, it is predicted, may grow to a $480 billion industry. In a best case scenario, humans and AI may come together in human-machine ‘superminds’, each playing to their strengths.

It is also often suggested that AI, like every major technological breakthrough of the past, shall create many new jobs we haven’t thought of yet. Human demand is insatiable, and as things grow cheaper through automation, the argument goes, the demand for other, harder to automate things may rise. Unfortunately, there is no certainty that older technological revolutions are analogous to AI. No matter how useful the machines of the past were, there was a wide gulf between their capabilities and those of a human being. That gulf is shrinking by the day. If there are interstices where humans will continue to outperform AI, they shall take time to locate.

The policy question

The capabilities of AI are a technological question. What we shall do in their wake, however, is a policy question.

While the industrial revolution, in the long run, brought about a marked improvement in human lives, those who lived through it barely saw its benefits. Several generations passed before real wages grew. Meanwhile, overnight, humans were ripped out of a way of life that they had maintained since the birth of agriculture. Cities, once a rarity, mushroomed to accommodate millions of new inhabitants. The structure of the family changed, perhaps permanently. Social and political cleavages erupted. New philosophies and systems of governance were evolved to cater to this newly transformed society. The shockwaves of this change are still making their way through parts of the world that have not yet completely industrialised.

Once again, we stand at the precipice of uncertainty, with new technology so powerful that we struggle to imagine a life with it. This time, however, we are better prepared, with more knowledge and better tools at our disposal. We need not repeat the same mistakes.

Top Stories of the Week

India makes moves on critical minerals

Last week, India became part of the Mineral Security Partnership (MSP), an exclusive grouping of thirteen countries, led by the United States, that aims to catalyse public and private investment in critical mineral supply chains.

Announced by the United States in June 2022, the grouping focuses on minerals such as Cobalt, Lithium, Nickel and rare earth elements, which are inputs in electric vehicles and advanced battery cells. Through the partnership, partners aim to secure a stable supply of these materials for their economies. MSP partners also profess to maintain high environmental and social governance standards in the mining, processing and recycling of such minerals.

The partnership, it is believed, is intended to act as a counter-weight against China, which has slowly cemented a hegemony over global minerals supply. This has made much of the world reliant on China for industries that, as we move towards decarbonisation, shall be increasingly important. With such reliance comes an increased risk of disruption in supply chains. These industries also hold growing salience for India, which is trying to seed EV and battery manufacturing industries domestically.

Under the MSP, countries develop lists of minerals that they consider critical, depending on the needs of their economy and their strategic assessment of supply risks. Accordingly, alongside its membership to the grouping, India has announced a list of 30 critical minerals.

Immediately following India’s entry into the MSP, it also inked a partnership with Australia for investment into critical minerals. With this, India shall source minerals that are processed in Australia - which produces half the world’s lithium, and is the world’s second largest producer of cobalt and fourth-largest producer of rare earths. To begin with, five projects - two for lithium and three for cobalt - have been marked for detailed due diligence.

Power Ministry’s new tariff system

India has witnessed a significant development in its power sector this week - the Government of India has introduced two changes to the prevailing power tariff system, which will enable consumers to save up to 20% in their power bills. These changes were brought through amendments to the Electricity (Rights of Consumers) Rules, 2020, notified by the Ministry of Power on June 14, 2023.

The first change is the introduction of a time-of-day (ToD) tariff, which means that consumers will be charged different rates for electricity depending on the time of day that they use it. During peak hours, which are typically the hours between 6pm and 10pm, the tariff will be higher than during solar hours, which are typically the hours between 10am and 4pm. This means that consumers who can shift their energy usage to solar hours or off-peak hours can save money on their bills. The ToD tariffs will be applicable from April 1, 2024 for commercial and industrial consumers having a maximum demand of 10 kW and above. For all other consumers except agricultural consumers, the ToD tariffs will be applicable from April 1, 2025.

The second change is the simplification of the rules for smart metering. Smart meters are meters that can measure electricity usage in real time, which allows consumers to track their usage and also identify areas where they can conserve energy. The simplification of the rules for smart metering will make it easier for consumers to get smart meters installed, helping them save money on their bills.

A Few Good Reads

A collection of prominent global leaders, including Joe Biden, Emmanuel Macron, Fumio Kishida, Rishi Sunak, Lula Da Silva, Ursula Von Der Leyen and many more, outline their vision of a climate transition that leaves nobody behind.
Rahul Verma shares a rare, balanced take on the health of India’s democracy.
Ever found yourself wondering why things feel so much worse? Why ethics and morality seem to fade away steadily; why the ‘age of kindness’ seems to have passed? Well, it might all be an illusion, writes Dr. Adam Mastroianni.
Pranay Kotasthane looks at what India can do to combat its brain drain problem.
Check out The Economist’s crony-capitalism index.

The Aakhya Weekly