A piece written in a hurry so I could put words on what I’m thinking right now, after a few days of trying ChatGPT-4. What follows may sound dramatic, but I don’t see how it could be otherwise.

Beta

I had tested GPT-3 quite a bit on fairly varied subjects: content writing, a bit of code, and even if I had been impressed, my use of it was really irregular.

OpenAI released the GPT-4 beta last week. I was obviously hyped by the demo videos and decided to get ChatGPT+ “just to see”.

And then, the shock. GPT-3 looks like a relic of the past.

I got about 5/6 days of work done in barely 1 day on technical tasks, including some I only had a vague knowledge of. I tested a lot of things: writing unit tests, writing tests, creating SQL queries, documentation, refactoring, code reviews.

Impacts on the IT sector - massacre

In a few words: it’s going to be a massacre. GPT-4 is going to multiply developer productivity tenfold, which, in itself, is an excellent thing. I don’t know by what factor, but a 1.25 or even a 1.5 doesn’t seem crazy to me: writing tests, comments, code review, stackoverflow help. (Microsoft is communicating a productivity of 1.55 on Github Copilot X)

You don’t believe me? Try it over a stretch of time, and not with prompts of the “go on, make me a React component” kind. It’s a bit more complicated than that.

What can happen? Well, OpenAI has solved the worldwide developer shortage. When every developer sees their productivity rise by a factor of 1.5, we’ll be able to lower the number of developers needed by a factor of 1.5.

What is going to happen to all those developers left on the sidelines? Expect a massive drop in salaries, by simple market effect, the same one that has been sending those same salaries soaring for 10 years now.

On top of that, the drastic slowdown of the current economy and the many cuts at tech companies that were overstaffed.

“Yes but not every developer is going to want to use it”

Sure, but those developers are going to have to raise their productivity by a factor of 1.5 compared with their colleagues/competitors who use an AI in their day-to-day, if they want to stay employable.

“Yes but not everyone is concerned, the experts won’t be touched”

Yes, I pretty much agree with that: the experts will gain the same productivity as everyone else.

“Not every company is going to use it, to protect their data”

I think every company, just like developers, will have to use it. It’s a matter of life or death, because their competitors will use it.

“Whoa, anyway, adoption is going to take a long time”

That’s actually what we’d need, but it’s going to be extremely fast, given the performance.

I think we’re going to see an impact equivalent to the arrival of the internet, but much faster and more brutal. Where the internet took a decade to change things, I think a few months will be enough (yes, I’m not talking in years).

Society is absolutely not ready for the shock that’s coming.

I’ll talk about the opportunities another time. For now it’s: Brace for impact

EDIT of 22/03/2023 Microsoft announces Github Copilot X based on GPT-4 with an announced productivity of 1.55. I wasn’t far off, and it’s better than what I had estimated. Coupled with VS Code, it’s the ultimate weapon, because GPT-4 will be able to look at project trees, make architecture recommendations, etc., not to mention the time saved switching from one window to another. It’s better than I imagined. And on price, even 1000 € a month per user would be profitable for a company. WE ARE FUCKED.