The hype was huge when it was released, and the early beta testers were showing some amazing (and cherry picked) demos, most famously the ability to write working React code. But since then, I've not seen much...
The hype was huge when it was released, and the early beta testers were showing some amazing (and cherry picked) demos, most famously the ability to write working React code. But since then, I've not seen much...
It’s increasingly difficult to find product reviews with search engines.
Massive auto generated content farms take a product name and add loads of AI-generated filler text. Pop in a bunch of banner ads and an affiliate link and they have huge economic incentive to scale these operations.
I’m very pessimistic about the direction the internet is going these days. The AI crisis isn’t going to be sentient AI trying to kill us, it’s going to be a flood of noise over knowledge.
https://mobile.twitter.com/moo9000/status/145873329934659174...
Human thought on the other hand has some sort of undefinable entropic-value that AI to-date is missing, a Human can produce a "good idea", whereas an AI produces a bunch of potential continuations of a stings of text and selects randomly amongst them (or, even better, a Human selects from them).
Unfortunately the advertising game mixes up the incentives and flips the equation so that the purpose of communication isn't to share a "good idea" as efficiently as possible, but rather to keep eyeballs on your website for as long as possible in the hope some flashy banner ad will distract your user and you'll get your $0.02 for them abandoning your page, likely unfinished. AI will (and already does) excel at this sort of task, but it's the kind of task that ought to have no value whatsoever.
Luckily we have increasingly sophisticated summarization-AI to go from the filler-AI generated crap back down to a couple of bullet points, but at that point you've invested millions of dollars, researcher-hours, engineer-hours, compute-hours, etc, to make the worst text-compression utility of all time.
We'll need another GPT-3 bot to detect the GPT-3 bots.
I navigate to the line where I believe "the fix should go here", and a few characters in, copilot is filling up the lines. 80% of the time it is non-compilable, but nearly 50% of the time it's close to the fix I was going to put in. It's then just a matter of me fixing much simpler errors and bugs in the copilot-suggested LOC.
I have found that I get far less distracted from writing bugfixes once I start looking at the code. I'm not going to let copilot push commits to PROD anytime soon, but it's like having a really smart intern who doesn't really know exactly what I'm trying to solve but has a decent idea, pair programming with me.
So it's not like these AI tools will replace me yet, but they are certainly living up to the goal of "copilot".
I mean - that is technically feasible.
That's barely scratching the surface of the AI-generated erotica scene, it's pretty wild.
Funnily enough, so long as you aren't trying for porn it does a much better job than AI Dungeon did of staying away from it. It lives up to its name — it's an excellent cowriter for all forms of fiction.
It's the exact opposite, in just about every possible way. Including, I'm afraid, model generality -- it's tuned for fiction, and nothing else, but it's very good at that.