The AI Value Paradox

I often talk about computers in this blog. And unless someone faxed this article to you, you are reading this on a computer. A computer is a marvellous machine. Through the history of computing, the early pioneers were immediately aware of the abilities and opportunities of the computer.

In 1950, Alan Turing proposed the Turing test to test if a machine could convince someone it were human. Colossus, the primordial digital computer, had only cracked the German Enigma cipher at Bletchley Park a few years prior. Throughout Turing’s tragically short lifespan, no computer existed that could come close to passing the Turing test. But even in those early days, it was easy to imagine they would.

It was about 66 years after Turing’s death that OpenAI released Chat GPT. If you have used Chat GPT before, you know it can carry out very convincing conversations. Certain models have passed the Turing test. In order to defeat the Germans, the code-breaking computers at Bletchley Park had to be state secrets. Now tools such as Chat GPT are very much a public concern. The only person not aware of AI is perhaps my grandma, who has never operated a computer and isn’t aware of the Internet. She’s so lucky.

We all know the story of the computer revolution. The birth of the PC, the Internet, the iPhone, et cetera. I won’t bore you with the details. During the revolution, early innovators and investors saw massive growth and returns. If you were part of Silicon Valley in the 1980s, and you were smart, you had an amazing opportunity to make money.

But the revolution was quiet. It was deliberative. It was nerdy. Moreover, it was not widely accessible to everyone. The world was a fertile landscape for digital technology to take root. It was a fantastic environment for investors and engineers, as well as actual human beings.

Now the landscape looks completely different. Technology companies dominate our world. Everybody knows it. But people still want a piece of the pie. This is the normal, healthy economy at work. If a lucrative work or investment opportunity exists, people will flock to it.

Innovation is the operative word. If you don’t build something new, you won’t become a grow company. The technology industry cannot operate like the sleepy electricity and gas utilities, which grow at the pace of people and geography. Nor can they be like the financial sector, growing at the pace of handshakes and relationships. They needed to grow at the speed of light, and shoot to the moon.

But the tech industry is now overexposed and diluted. All eyes are on it. And the expectations are high. There is nothing that can grow, scale, and make money quite like hardware and especially software. Today, the eyes focus on generative AI. Generative AI is a broad term, encompassing chat-bots, image generators, and other applications. But people just call it AI.

When someone shows you a picture they made with AI, what do you think of it? In the early days of generative AI, the response was often wow, a computer made that? Nowadays, you would shrug when someone shows you their AI-generated image. After all, anyone can type what they want into the AI website’s textbox and get it to fart out a glossy PNG of whatever they are thinking.

If that someone still wanted to impress you, they might build their own AI image generator. Then you’d be impressed: wow, you made that? Producing machine learning software is not an easy feat. You would be proud and excited if you pulled it off. But as soon as you switched it on, and it started making images and writing text, your AI would generate more shrugs than useful data. That’s cool, people would say, but there are heaps of other websites where I could do that.

AI will automate a swathe of processes people have to carry out. This will save time and money. AI will be universally applicable across industries. Let’s roughly consider the business factors. Conventionally, a business increases profits by either reducing costs or increasing revenue. To increase revenue, companies can apply AI to produce more, or better, products and services. Likewise, they can automate work to reduce costs. Whatever form AI takes, businesses only really care about revenue and cost.

Workers are already excited for the AI revolution.

The economy is full of equilibria. In a free market, businesses compete with each other. In theory, competitive equilibrium means businesses reduce prices to the lowest level they can (while remaining profitable). Remember that while businesses compete to lower prices, they want to maximise profits. If automation reduces labour cost, and prices stay fixed, then profits increase. If automation reduces price, but revenue increases due to higher sales, then profit also increases.

Your business will make money from AI if automation gives you an advantage over others. When computers and the Internet were relatively unknown, it took effort and investment to acquire a computer and “get online”. If you pulled it off well, profits would increase. Now the same digital technology, which got everybody online, has made the next revolution accessible and everywhere. You won’t be the first to take advantage of AI technology. Instead, you and your competitors have FOMO. Everything equalises. Competitive advantage from AI is lost.

Three authors explore the loss of competitive advantage in a Frontiers of Psychology article. They call it commoditisation. Commodities are materials such as coal, petrol, and electricity. These are important, but not particularly special, and certainly not scarce. AI is neither scarce. You can easily use the AI tools that someone else made. You don’t need to put in the effort to produce your own. Instead of being a business benefit, AI becomes the status quo. It is another operation that businesses need to plan around and manage. With that, the benefits of AI not only disappear, but become another business risk.

If Mr. Turing could time-travel to 2025, he would marvel at the AI models that could pass his then-hypothetical test. But we don’t. Instead, after acknowledging AI’s capabilities, we realise the disruption it poses to many jobs. We face the loss of the human touch from AI-generated content and “art”. Generative AI can “generate” a lot, but we may lose so much as well.

Ordinary people and companies may need to adapt to and implement AI to not be left behind. But they are unlikely to realise a monetary benefit through automation. Nonetheless, early investors are already seeing returns. Nvidia has seen a massive increase in its share price in the last five years. They produce computer processors that generative AI needs. Having reached the bottom of Goldrush Mountain early on, they get to sell the shovels. Whether the rest of us are hyped for AI, or begrudgingly accept it, we need the shovel. And we still need to climb the mountain.

I use Chat GPT to edit and check these articles. Because I value you, I don’t ask it to rewrite anything or do the important work for me. After all, anyone could do that. And then my articles would not be worth it. But I cannot deny how powerful generative AI is. And it is poised to give this power to everyone. If something is valuable, and everyone gets it, it ceases to be so valuable. But I do not know the economics of AI. I do not know what the future holds for us and AI. And I do not know what it will all mean for me. But I do know a quote from The Incredibles:

When I’m old and I’ve had my fun, I’ll sell my inventions, so that everyone can be super – EVERYONE can be super! And when everyone’s super, no one will be.

What a paradox.