The AI Development Era - GPT-4, Bing Chat, and Bard
It's been about a week since GPT-4 launched, and I've been using it daily.
My impression so far: "Slow, but good."
Recently I've been using Bing Chat alongside the GPT-4 model.
For actual data research purposes, Bing is definitely better.
But GPT-4 is overwhelmingly superior for development, brainstorming, documentation, and personal learning.
Earlier this morning, Google's Bard launched in limited beta in the US and UK.
It seems to occupy a position somewhere between Bing and ChatGPT, but for now it still feels somewhat ambiguous.
Above all, I've realized that using these models with simple prompts and parameter settings is far more powerful than the awkward fine-tuning approach.
Because of this, I expect AI-powered content will become increasingly accessible going forward, and an even wider variety of products will flood the market.
And while there's constant talk about developers disappearing, rather than developers vanishing entirely, I think the dedicated programming developer field will expand into what's called "prompt engineering" — though frankly, pure coders might indeed become obsolete.