Having a really good idea is hard
You’re probably aware that the Metaverse is now dead (have a read of Mark Ritson’s analysis of its demise here).
As I never tire of reminding people, Mark Zuckerberg only had one ‘good’ idea: steal someone else’s good idea (when I say ‘good’ I mean good for his bank balance, not for the world or the many people who live in it). Unfortunately this idea proved to be so incredibly successful he (and others) believed he was the kind of guy who is capable of producing more good ideas.
Reader, I’m afraid he did not produce more good ideas. In fact, he had a really bad one. The Metaverse nonsense cost $80bn, which now means it wasted $80bn. Imagine having an idea that cost that amount of money for zero return. You could even argue that it has had a negative effect, with the word ‘meta’ featuring all over his company’s products as a sardonic reminder of how crap his idea ended up being.
But successful ideas of this scale are so rare that that chances of anyone having more than one are virtually zero. Most are iterations of ideas that were already good but needed the odd improvement. Even Facebook was basically Friends Reunited/Friendster with some improvements. The iPod was an improvement on the MP3 players of the time, and the iPhone, iPad and Apple Watch were all improved versions of existing products. The app store was a great invention, but the people responsible were not aware of that at the time.
I remember being at Media Arts Lab just after the launch of the iPad 2. It seems funny to think back to that moment and recall that there was a loud, dissatisfied clamour from people saying ‘What’s next?’ and ‘Why hasn’t Apple produced another amazing new thing in the twelve months since the iPad?’. Of course, that clamour died down, but there seemed to be an expectation that, with three world-changing products in the bag, more would be on the way. Aside from its various earphones/earbuds, Apple has yet to launch a new ‘game-changer’, and their virtual reality headset has predictably gone the way of the Metaverse. That’s fifteen years of the most innovative and successful company in the world producing nothing new.
In the last 5-10 years we’ve seen blockchain, Web3, NFTs and crypto currencies hailed as the next big thing, only to see them fail to one degree or another, taking billions of dollars of investment down with them. Sure, they still exist in one form or another, but they hardly feel like the Next Big Thing they were promised to be.
Which leads us to generative AI. Now that we’re three or four years into the brave new world of Chat GPT, I think it’s fair to consider the extent to which it has become the success it promised to be.
My personal usage of it seems to be pretty similar to how it’s always been: a bit of research, a few paragraphs to fill up decks, the kind of creative partnership I’d get from a so-so junior creative who’s really good at spelling – not bad but little improvement. The imagery production facility is better, but my usage of that offering hasn’t really increased.
I understand that many companies are making greater use of generative AI but the pace of that progress remains uneven. If, like me, you expected to be in it up to your eyeballs by now, you might be surprised at how similar your usage is to that of, say, 2024. In addition, only 16.3% of the word uses it, and that number went up by just 1% last year.
The innovations of generative AI seem to have been a bit of a mixed bag, with no real leap forwards despite many such promises. I’ve been hearing the term ‘Agentic AI’ for over a year, but no one seems entirely able to present its use case in a way that makes it seem compelling, let alone inevitable. In addition, the spasms of delight that accompanied the launch of Sora’s video generation have fallen to silence as it has now been shuttered and its billion-dollar deal with Disney canceled. Open AI has also canceled its ‘erotic chatbot’, and shifted all its resources to using its AI to improve its AI. Add in the post Iran-war increase in the cost of energy, something AI needs an enormous amount of, and you get the impression of a rocket ship that has very much plateaued.
This underwhelming progress has made me wonder if the problem has been a an unfortunate collision of expectations and reality. Our early excitement at what generative AI could bring was based on initial usage and word of mouth. It was easy to see how its more complete search facility might destroy Google. Perhaps it would be cranking out perfect books or movies based on any half-arsed idea that might pop into our minds. At the very least it should have been saving us many hours of drudgery by creating perfect presentation decks in milliseconds.
Of course it has offered none of those things, leaving the early promise to drift away, with no new compelling usage cases to impress the majority of its users.
I suspect that the constant drumbeat of ‘By 2027 it’ll be able to do X’ and by 2028 ‘X number of people will be doing Y’ has now become something of an Emperor’s New Clothes scenario. With every promise that turns to dust, skepticism grows, as does the pressure for a clear game-changing development that will keep us impressed enough to be confident of even more greatness in the future.
It’s starting to feel more like a facility in search of a compelling use case, one that we can understand quickly, not something vague about ‘agents talking to agents’ that sounds more off-putting and scary than exciting and clear.
Perhaps generative AI has had its one ‘really good idea’ and is now scrabbling around in its equivalent of the Metaverse, hoping that it digs up something that will capture our imaginations again.
At the moment it’s crying wolf too often, which means that pretty soon we’ll stop listening.