With the arrival of ChatGPT the workaday think-piece journalists are out in force with their hot takes about how this is the end of search. Or the end of high school english. Or the end of, well, all sorts of things. And, you know, maybe it is.

But, then again, maybe it isn't. Maybe it's just a refinement of what we already had.

To give an example: the gasoline, or internal combustion, engine. Was it ground-breaking? Of course. Did it replace steam engines? Absolutely yes. In all sorts of applications. But did it do away with trains (that were, I would note, running at that point on steam engines)? Or boats? Absolutely not.

Trains and boats are definitely still things we have, and use, and are the most efficient ways to go about the tasks we put them to. And the ICE also opened up automobiles in a practical way that made it more wide-spread. But I would note that the ICE was a feature of these products. It was not the product itself. Outside of a few crazies, an engine is not itself the product.

I'm going to focus on what I see as the most widely-cited potential disruption: search. A lot of people are asking 'Will ChatGPT replace search?' And, of course, 'Will this kill Google?' (Disclosure: I currently work for Google, though in Cloud and not Search.) I think that those asking this question are looking for status (or a paycheck) from think-piecing.

Consider that the internet is already filled with gobs of misinformation. And consider that these machine-learning AIs have to learn from a corpus of work. Worse, that they are going to do the most-average job they can, by the nature of ML - they can only look for the most likely thing, which is going to be the center of all the signals they have. Alone, this form of AI will do worse than average, at best. And Google didn't kill every other search engine product by being slightly better (never mind slightly worse) than every other search engine. It was a quantum jump forward as a product. It was simpler. It didn't sell you anything (RIP). And it gave you wildly better results. The search engine product was a quantum leap - one fueled by innovations in web crawling, indexing and product UX.

What the internal combustion engine needs is a whole bunch of other features to become a product. Rubber wheels. Transmissions. Shock absorbers. Batteries. ChatGPT needs a very accurate corpus, at the very least. A corpus which, almost certainly, will be sourced with, well, search. ('Search' as we understand it today.)

I think it is important to understand that ChatGPT is a big step forward - in our interface with information. It presents a new way we can process and leverage that information - so far in giving natural-language responses that sound authoritative and are divorced from source material. But this will be a feature of many (often dystopian) products, not the product itself. What it offers us right now is as easily obtained in a more accurate, less biased form - which makes it a feature, and not the world-disruptor it is being posed as.