no, non c'è pericolo. è invece interessante osservare come le medesime tecniche generative producano risultati soddisfacenti (almeno sotto qualche aspetto) se applicate al linguaggio naturale, ma assolutamente casuali (29% di accuratezza è nulla) se applicate al codice. il fatto è che per il codice la semantica non è un 'optional' :-)
G. On Mon, 26 Jul 2021 at 09:10, maurizio lana <[email protected]> wrote: > in questo ambito è possibile che la moneta cattiva scacci quella buona? > cioè che il codice scritto dai sistemi di AI, anche se crappy, soppianti > quello buono /ottimo prodotto da analisti e programmatori capaci? > se è possibile, si può fare qualcosa cosa per evitarlo? > > nel senso che se già è così difficile [eufemismo] che sistemi di AI > prodotti da soggetti umani operino in modo eticamente e culturalmente > soddisfacente, c'è da chiedersi come possano operare dei crappy software > prodotti da sistemi di AI imperfetti [eufemismo] > Maurizio > > > > Il 25/07/21 12:00, [email protected] ha scritto: > > Date: Sun, 25 Jul 2021 09:52:35 +0000 > > From: Alberto Cammozzo <[email protected]> > > To: Nexa <[email protected]> > > Subject: [nexa] Is GitHub Copilot a blessing, or a curse? · fast.ai > > Message-ID: <[email protected]> > > Content-Type: text/plain; charset="utf-8" > > > > <https://www.fast.ai/2021/07/19/copilot> > > > > [...] > > According to OpenAI’s paper, Codex only gives the correct answer 29% of > the time. And, as we’ve seen, the code it writes is generally poorly > refactored and fails to take full advantage of existing solutions (even > when they’re in Python’s standard library). > > > > Copilot has read GitHub’s entire public code archive, consisting of tens > of millions of repositories, including code from many of the world’s best > programmers. Given this, why does Copilot write such crappy code? > > > > The reason is because of how language models work. They show how, on > average, most people write. They don’t have any sense of what’s correct or > what’s good. Most code on GitHub is (by software standards) pretty old, and > (by definition) written by average programmers. Copilot spits out it’s best > guess as to what those programmers might write if they were writing the > same file that you are. OpenAI discuss this in their Codex paper: > > > > “As with other large language models trained on a next-token > prediction objective, Codex will generate code that is as similar as > possible to its training distribution. One consequence of this is that such > models may do things that are unhelpful for the user” > > > > One important way that Copilot is worse than those average programmers > is that it doesn’t even try to compile the code or check that it works or > consider whether it actually does what the docs say it should do. Also, > Codex was not trained on code created in the last year or two, so it’s > entirely missing recent versions, libraries, and language features. For > instance, prompting it to create fastai code results only in proposals that > use the v1 API, rather than v2, which was released around a year ago. > > > > Complaining about the quality of the code written by Copilot feels a bit > like coming across a talking dog, and complaining about its diction. The > fact that it’s talking at all is impressive enough! > > > > Let’s be clear: The fact that Copilot (and Codex) writes > reasonable-looking code is an amazing achievement. From a machine learning > and language synthesis research point of view, it’s a big step forward. > > > > But we also need to be clear that reasonable-looking code that doesn’t > work, doesn’t check edge cases, and uses obsolete methods, and is verbose > and creates technical debt, can be a big problem. > > > > > > > ------------------------------------------------------------------------ > > the knowledge gap between rich and poor is widening > witten, bainbridge, nichols, how to build a digital library > > ------------------------------------------------------------------------ > Maurizio Lana - 347 7370925 > > _______________________________________________ > nexa mailing list > [email protected] > https://server-nexa.polito.it/cgi-bin/mailman/listinfo/nexa >
_______________________________________________ nexa mailing list [email protected] https://server-nexa.polito.it/cgi-bin/mailman/listinfo/nexa
