What are you on about? This is exactly what LLMs like GPT-3 or GPT-4 can and will solve. It just takes some time. But the capability to understand, reason about and execute via API calls such simple instructions has absolutely been demonstrated. Getting to a shipped product takes longer of course.
Would you mind sharing a few examples of these demonstrations? I’m genuinely very interested.
What I’ve seen so far is that they are so flaky that can’t be relied on in general. And that’s exactly my point, that the demos are cute, but can’t be relied on for ordinary day-to-day tasks.
That’s fine, progress has been great, and hard problems take time. But see the comments and people are freaking out that “AGI” is around the corner and smarter than humans and other nonsense claims.
It seems like a great example. This can be done by just chaining together existing models and really doesn't take much sophistication. Give it some time and we will get out AI assistants. After all, computers understand language now. This is plainly what has happened over the last few years.
