No technology trend goes straight up. There are always crashes along the way, as subsets of the trend fall behind, while others prove their worth.
This was true with the Internet. Search engine companies were falling even before Google was created. It’s why Wall Street wanted Yahoo to become a “portal.” (Wall Street knows nothing.)
This will be true for A.I. as well. It’s starting right now.
The “consumer side” will crash first. Things like ChatGPT were fun for a while, but the mass market is deciding it’s more like Pokemon Go than “the Google.”
For any technology to gain legs, it must prove its utility. It must create productivity. It must not only do things that weren’t available before, but it must make money doing it, and make doing other things easier.
In the case of things like MidJourney and Discord, it’s unclear what those productive, profitable things are. It’s easy to see where Adobe will make its money, but Microsoft is overpricing, doubling the cost of what was once Office if you want the A.I. features. That’s a great way to limit take-up, limit profitability, and give everyone else a chance to catch up.
The beneficiary of this, I’m sorry to say, will be the Artists formerly known as Facebook. Meta Platforms is releasing its language models as open source. Meta is the dominant social media platform in the developing world. Millions of programmers, for whom English is difficult, are getting the tools to appear erudite. They’re getting tools to make creating and documenting their code easier. They’re getting tools to give their proposals developed-world quality.
Basing Threads on the open source Activity Pub may sound brilliant, but it’s going to cost Meta. Every nation, and every sub-group within every nation, will now be able to create a Mastodon instance compatible with Threads, gaining the benefits of separation without losing access, or belonging, to the larger sphere. Over time this is going to further Balkanize the Web, because governments will be able to control these smaller instances, but those are the breaks. It’s been 30 years since the Web was spun. We know you’re a dog.
Somewhere in here will be a reshuffling, away from basic enablement like cloud and A.I. “operating systems,” toward applications and solutions. Once the market gets downwind of this, expect money to move, away from the Cloud Czars, and down the value chain.
