Socrates and the rise of artificial intelligence

Five years ago, we were in the midst of the Covid-19 pandemic. Apart from the SARS-CoV-2 virus itself, the other thing that spread as fast was the discussion that went with it. Fuelled by social and traditional media, we underwent crash courses in epidemiology, learning about curves that needed to be flattened, new variants named after Greek letters and terms like cytokine storms (there were also conspiracy theories, but we won’t go into those here).

Something similar is going on now, as we enter the age of artificial intelligence (AI). A mix of existential fear and curiosity is fuelling a new discussion and the emergence of instant experts, all trying to answer the question: What Does This All Mean?

We don’t profess to have all the answers, but, in what follows, we hope to at least set the scene with some of the key questions that AI is posing.

Is AI making us stupid (and lazy)?

A colleague recently highlighted a study of the behaviours and brain activity of a group of subjects by researchers at the Massachusetts Institute of Technology. The subjects were divided into three groups and asked to write SATs (a standard group of tests for university entrance in the US) using, respectively, the generative AI platform ChatGPT, Google search, and no external aid at all.

Using electroencephalograms (EEGs) to monitor the brain activity of the subjects, the researchers found that of the three groups, the ChatGPT users had the lowest brain engagement and that, over three months of doing the study, they became lazier, often simply cutting and pasting the content from ChatGPT without checking it.

The group using the search engine and, especially, the group using no external tools at all, showed higher levels of neural activity, were more engaged in their work and expressed higher satisfaction with their output.

The research paper (which has not yet been peer reviewed) suggests that large language models like ChatGPT could harm learning and, by extension, our sense of satisfaction on completing a task.

If the AI tools do more of the work, though, how will people learn and grow, especially young people? A big part of working with AI, I believe, especially generative platforms, is that you need to apply a critical eye to spot not just the errors (hallucinations as they are called), but also the nuances that may not be technically incorrect but which may miss key points or aspects.

Spotting these requires experience, often of the unglamorous kind. But this experience usually only comes from doing the things that AI now does, seemingly effortlessly. How can younger employees acquire the experience they need if they aren’t getting their hands dirty doing the often-laborious work that AI is supposed to do for us?

Girl you know it’s not true – blame it on the AI

And what about promoting creativity? AI draws on the material and patterns that are already out there, so is, by definition, uncreative. In a recent podcast by Irish economist David McWilliams, he noted that AI essentially looks backwards, rather than tries out new things. The AI reinforces that which is familiar.

True creativity needs risk-taking and boldness, though, or we become stuck in iterations of the same things for eternity.

In a related problem, McWilliams also referred to what the billionaire investor Mark Cuban called the “Milli Vanilli effect”. Milli Vanilli was a 1980s R&B duo that had hits like “Girl you know it’s true” and “Blame it on the rain”.

It was subsequently discovered that they had lip-synced all their songs, which had all been sung by session singers. Following a public outcry, they even had their Grammy Award revoked.

Milli Vanilli may never have been outed were it not for a faulty backing track during a (supposedly) live performance for MTV. Nowadays, as the number of AI-generated videos out there grows and the deepfakes become more realistic, it’s becoming more and more difficult to tell the fake from the real thing.

This creates two dangers for society. One is the potential to whip up extreme public rage in an already fractious world, and two, once people realise that there is so much fake media out there, it will break down trust in public media and institutions, with severe consequences for our culture and possibly for world peace.

Could AI destroy jobs?

Another concern raised by the rise of AI is job security. Goldman Sachs recently predicted that up to 50% of current jobs could be fully automated by 2045.

Twenty years from now is bad enough, but the issues seem to be manifesting already. Big tech employers like Microsoft, Intel and the big consulting firms have already started to either reduce headcount in key areas or slow down new recruitment in favour of AI tools.

According to the Financial Times (quoting data from job search engine Adzuna), the number of listings for entry-level graduate roles has fallen by two-thirds since 2022 in the UK and by 43% in the US.

Meanwhile, AI firm Anthropic has warned that half the jobs for under 30s in administration, management and tech could be gone within five years.

Historically, technological developments have always had an impact on jobs, but society has adapted with new careers, even though the impact has often been traumatic in the short term.  

Yet the AI boom seems to be happening at a much faster pace than previous disruptive digital technologies. For example, ChatGPT reached one million users within a week of its launch in November 2022 and 100 million users within two months. Compare this with the four-and-a-half years it took Facebook to reach 100 million users and Instagram’s 2.3 years. This implies a rapid displacement in the workforce in the coming years.

It’s not clear how quickly people will adapt to these rapid changes, and one fear is that AI-driven layoffs could even trigger an economic downturn or exacerbate inequality. More on this further below.

Is AI bad for the environment?

Mistral AI, an AI tool primarily for software developers, recently released an audited review of its water, energy and material consumption over the past 18 months.

It found that each marginal query uses 45ml of water (the equivalent of nine teaspoons of water) and releases one gram of carbon, based on how electricity is generated at its data centres. 

Why Socrates didn’t write things down

So there you have it: AI will make us lazy, stupid, uncreative, angry, apathetic and without work. And it’s bad for the environment.

Or maybe not. Is there a more optimistic take on things?

In times of major technological change, it’s a good time to look back in history at how society initially responded to major technological advances and how behaviours changed over time.

One of the best examples is one of the earliest – writing. The Greek philosopher Socrates (c. 470 BC to 399 BC) distrusted the notion of writing things down, saying that it diminished the need for memory, and led to a shallow understanding of things.

This sounds preposterous to us today.  The written word has enriched us, giving access, through books, libraries and the internet (among other media), to vast stores of knowledge (and nonsense too, it must be said, but that’s a subject for another day).

It’s worth noting that we wouldn’t have known about Socrates’ views on writing if they hadn’t been written down (in Plato’s Phaedrus, in 370 BC).

There are other good examples from more recent times, such as the calculator, the spreadsheet and accounting software. Jobs in accounting and finance haven’t disappeared and these technologies may even have increased the number of job opportunities.

AI or nay?

Many believe that AI will drive the global economy to new heights. A recent article in The Economist suggests that AI could boost GDP growth significantly, with some “evangelists of Silicon Valley” putting the increase at between 20% and 30% a year.

Even a much lower number would have a profound effect on productivity and the market for goods, services and labour. As noted above, there might not be an immediate boost in job growth, but this will hopefully come in time, disproving once again the “lump of labour fallacy”.

Perhaps we will need intervention by lawmakers and regulators to help smooth the transition and to ensure AI adoption is implemented ethically and in a way that looks after those whose livelihoods are under threat.

Hopefully, the technology can also be put to use in solving some of its own problems, such as ways to make data centres more energy efficient or in identifying deepfakes.

At the heart of our future as we tackle this new world is to remember the things that make us human. Here we refer to qualities such as curiosity, scepticism and creativity – in short, our capacity to say “what about this?” or “what if we try this?” These are the qualities that have driven many of our advances and are crucial to being able to thrive in an age of AI.