3 ms·
Does the library work? Also, this is an incredible achievement, but I don't understand this statement: > This wasn't something that existed, it wasn't regurgi
by burkaman 2y ago
Does the library work?
Also, this is an incredible achievement, but I don't understand this statement:
> This wasn't something that existed, it wasn't regurgitating knowledge from Stackoverflow. It was inventing/creating something new.
I mean you literally told it to copy an existing library. I understand the Haskell code is novel, but translation is not the same as inventing.
- echelon 2y ago> this is an incredible achievement If it was real, why aren't all the unicorns in this space worth $100B already?
- f1shy 2y agoWe could really make a valuation, considering that it can replace X % of developers… I think 100B is even low!
- echelon 2y ago> I think 100B is even low! It could be. It also depends on whether there is a moat around the tech or if we'll have a dozen fungible clones. But I'd certainly expect more signal if these models could do as advertised.
- bilekas 2y agoYeah, I have a problem with general statements like that ". It was inventing/creating something new." This simply is not the case as far as we know.. I'm yet to be convinced that anything under the hood is more than a clever predictive generator for now. It can appear to do really impressive things, but that's only while it's behaving correctly. All it takes is one bad line and the snowball effect of the problem is in motion. Lastly, it is true : > "the better the type system of the programming language and verbosity of compiler errors the harder LLM's go" The LLM has the best opportunity to succeed in this case. Nice to have but inventing, I doubt.