Hugging Face and ServiceNow Release a Free Code-Generating Model (techcrunch.com) 13
AI startup Hugging Face and ServiceNow Research, ServiceNow's R&D division, have released StarCoder, a free alternative to code-generating AI systems along the lines of GitHub's Copilot. From a report: Code-generating systems like DeepMind's AlphaCode; Amazon's CodeWhisperer; and OpenAI's Codex, which powers Copilot, provide a tantalizing glimpse at what's possible with AI within the realm of computer programming. Assuming the ethical, technical and legal issues are someday ironed out (and AI-powered coding tools don't cause more bugs and security exploits than they solve), they could cut development costs substantially while allowing coders to focus on more creative tasks.
According to a study from the University of Cambridge, at least half of developers' efforts are spent debugging and not actively programming, which costs the software industry an estimated $312 billion per year. But so far, only a handful of code-generating AI systems have been made freely available to the public -- reflecting the commercial incentives of the organizations building them (see: Replit). StarCoder, which by contrast is licensed to allow for royalty-free use by anyone, including corporations, was trained on over 80 programming languages as well as text from GitHub repositories, including documentation and programming notebooks. StarCoder integrates with Microsoft's Visual Studio Code code editor and, like OpenAI's ChatGPT, can follow basic instructions (e.g., "create an app UI") and answer questions about code.
According to a study from the University of Cambridge, at least half of developers' efforts are spent debugging and not actively programming, which costs the software industry an estimated $312 billion per year. But so far, only a handful of code-generating AI systems have been made freely available to the public -- reflecting the commercial incentives of the organizations building them (see: Replit). StarCoder, which by contrast is licensed to allow for royalty-free use by anyone, including corporations, was trained on over 80 programming languages as well as text from GitHub repositories, including documentation and programming notebooks. StarCoder integrates with Microsoft's Visual Studio Code code editor and, like OpenAI's ChatGPT, can follow basic instructions (e.g., "create an app UI") and answer questions about code.
Hugging Face? (Score:2)
Re: (Score:3)
Re: (Score:2)
It really was the Alien movies that have face huggers.
Yes, but that was a wimpy off world setup of Aliens that you could "nuke from orbit". The Black mirror version would likely involve the queen setting up in the Irish pub in your local mall.
Hugging Face and ServiceNow (Score:2)
Re: Co. names sound AI generated (Score:3)
AI gave me a catchy name for my new toilet tech discussion site: Flushdot
Fuck ServiceNow (Score:3)
Without a doubt, ServiceNow is the worst asset inventory software I have ever had the misfortune of being forced to use. It's ticketing portion barely qualifies as a ticketing system.
Anyone who ever buys this product should be investigated for kickbacks because I can't conceive of anyone willingly using this shithole product.
Re: (Score:2)
Without a doubt, ServiceNow is the worst asset inventory software I have ever had the misfortune of being forced to use. It's ticketing portion barely qualifies as a ticketing system.
Anyone who ever buys this product should be investigated for kickbacks because I can't conceive of anyone willingly using this shithole product.
I can confirm that ServiceNow is a lousy product based on proof shown to me by a now passed-on friend. My friend's employer did many "dodgy" things because the top-level management "approved of it".
Re: (Score:2)
A lot of enterprisey stuff is like that. I recently migrated our company off an "enterprise" time series database to good old trusty postgres (with the timescale extensions), and man what I found in that enterprise db was just astonishing. Threre where tables with multiple primary keys (not composite key, not primary and secondary, multiple *primary* keys), records that didnt fit the schema (Its a fixed schema db, not one of those smoothbrained mongo type things) Integer fields with text in them and so on.
I look forward to normals storing AI state backup (Score:2)
A lot of coding example I have seen, have been the magical examples of "make a website that does x, oh now I want to add a gradient background, oh now I want a from that takes an email".
That's great and all, but what happens six months later when you say "oh now I want a second form that stores data X" and the AI is like WTF you talking about man?
Then you feed it the source code and it tries to guess what to add where not understanding the semantic structure of what it itself generated.
I can see use of AI a
Shitty.JS (Score:2)
Re: (Score:2)
It is basically the last stand of "managers" that cannot admit that coding is actually hard.
Until it's perfect it isn't good enough (Score:2)
Until it's perfect, it's not good enough.
I've written code from scratch and I've updated the code of others. Debugging someone else's work when you don't understand what they were thinking as they wrote it gets exponentially more difficult the more code there is and the further they drifted from textbook examples. And you can't build a decent program from textbook examples copied verbatim.
If you let 'AI' write your code, you're not going to be 100% sure where to test, or what to test, and you're going to