AI’s impact will parallel that of the internet, not steam engine
It’s fashionable to use much older analogies for it than we should
and important facet, which is that “the youth unemployment rate increased with the level of education, with the highest among graduates.” Even here, the authors note that this unemployment rate has seen a decline since 2019 (actually since 2018).
They further point out that “unemployment in India was predominantly a problem among youth, especially youths with a secondary education level or higher.” The report then goes on to note its most cited and misunderstood statement: “In 2022, the share of unemployed youth in the total unemployed population was 82.9%,” which some political commentators have chosen to read as 82.9% of our educated youth are unemployed. When we recognize that aggregate unemployment affects a small proportion of the labour force, the mis-reading seems to be a deliberate case of ‘none so blind as those who will not see.’
In discussing youth unemployment, we note a second curious feature in their otherwise detailed analysis. The ILO-IHD report observes that the “incidence of unemployment was much greater… among younger youth than older ones.” It notes that in 2022, the unemployment rate was 13.2% for the age group 15-19, compared with 8.6% for the 25-29 age group. However, if we refer to its appendix table A4.2a, we would note that the age group 20-24 has the highest unemployment rate of 16.9%. Noting this difference is important because it highlights the transient character of youth unemployment. If we were to look at the next age band 30-34 (as is available in the PLFS report), we would see that the unemployment rate has further reduced, and is now close to the all-age group average. Even as the report’s authors include relevant data in annexures, the main report does not bring out that in each age group, the unemployment rate has seen consistent declines from 2019 onwards. Further, the gap between the unemployment rate in the age group 20-24 and 25-29 has also narrowed, suggesting that fewer people are staying unemployed, and for shorter periods of time. It is a pity that the authors of the report did not try to assess how long it takes people to get a job and the magnitude of improvement in this over the years.
The report also brings out that educated youth typically get better jobs. The econometric model used shows that higher levels of education yield better returns, and that there has been “increasing importance of higher levels of educational qualifications among youth in accessing higher paying jobs in the labour markets.” They also note that in 2022, there was a rise in the returns to youth with both informal and formal vocational training, indicating that the “government’s intensive efforts to expand skill training might have led to better returns for youth with vocational training.” These findings, when combined with the rising share of educated youth, bring out a key element of the country’s improving labour-market conditions seen over the last six years.
The ILO-IHD report deserves to be read without ideological blinkers to understand the transformations which have taken place in recent years in the Indian labour market.
Technologists have been doing it. Jamie Dimon just did it in his latest letter to shareholders. I’m referring to the way people are comparing the transformational impact of artificial intelligence (AI) to that of the steam engine. The metaphor has not only become a cliché; it paints an oversimplified picture of how this technology will reshape our lives.
To be fair to Dimon, CEO of JPMorgan Chase, his examples were drawn from a wider net: “Think the printing press, the steam engine, electricity, computing and the Internet, among others,” he wrote. But the effects of perfecting steam power pale in comparison with the changes that the next technological development will bring. The comparison is common. Microsoft chief technology officer Kevin Scott has said that it was the closest metaphor “to help understand what AI means for humanity.” Yes, harnessing steam pressure to run machinery and trains was pivotal to the Industrial Revolution, and yes, we are arguably in the midst of a metamorphosis with AI driving profound change.
For a start, AI’s impact will be far broader than that of the steam engine, transformed physical labour, manufacturing and transport. Today, AI models can generate ideas and art. Ad agencies are using them to brainstorm ideas and generate scripts and storyboards. This is an altogether different impact on decision-making and creativity, even personal identity and the way people socialize. Note the rise of AI chatbots like Character.ai, Replika and Kindroid, used for therapy, companionship and romance.
AI has also been adopted far more quickly than the steam engine. Thomas Newcomen’s first commercially successful engine in the early 1700s wasn’t improved on by James Watt until more than 60 years later; it would take another 150 years for steam power to be broadly adopted in manufacturing and railway locomotives. Contrast that with the way machine-learning algorithms have become prevalent in social media, retailing, logistics and more in just the last two decades. And the true catalyst, which gave rise to the latest era of Generative AI that conjures text, images, voice and videos, and tools like ChatGPT and Midjourney, was invented just seven years ago.
There are also big differences in the ethical and social implications of the steam engine versus AI. The former increased the rate of urbanization and the exploitation of human labour; the latter’s ethical challenges are more nuanced and arguably more insidious, relating to our personal privacy, surveillance, an erosion of human agency and creativity as well as potentially profound effects on personal freedoms.
Finally, nobody in their right mind ever worried about the steam engine going rogue and destroying civilization—but tens of millions of dollars are being spent to research just that possibility for AI.
Analogies are wonderful, but they should be picked with care when language has the power to shape opinion. During the Gulf War, for instance, the use of terms like ‘smart weapons’ implied a bloodless conflict with precise targeting that wasn’t actually possible.
Similarly, the discourse around artificial intelligence teems with illusory terms like ‘intelligence’ (machines aren’t intelligent), ‘neural networks’ (they don’t have brains) and ‘machine learning’ (they don’t understand and experience things in the way humans do), all helping to personify AI systems as something more human than they are in reality. The steam engine, whose harmful effects on human labour are a distant memory and which mostly brings to mind positive transformation, also doesn’t give us the full picture of AI’s repercussions. Instead, it gives us a rose-tinted view of the future.
Here’s a better analogy: The internet. Not only did it seamlessly weave itself into the fabric of daily life, just as AI is doing, it evolved rapidly from its inception, revolutionizing media and the way we socialize and communicate. The ethical problems it created around privacy, surveillance and misinformation are rearing their heads once again with AI, as are those around the concentration of power among a handful of Silicon Valley gatekeepers such as Alphabet’s Google, Meta Platforms, Amazon.com and Apple Inc.
Comparing AI to the internet offers a broader and more nuanced understanding of its potential impacts, not to mention one that hasn’t been softened by the passage of time. We can all still feel both the positive and negative side effects of the world wide web on our lives.
Overall, it’s a better comparison than the steam engine—as are the printing press, electricity and any other revolutionary inventions from the days of yore. But if you’re going to draw just one parallel from history for the potential of AI, it’s best to stick with the internet.