What is the point of converting this software?
-
@drwho Only until stuff breaks.
-
@drwho Never works, but they are welcome to try.
-
@samir Just let the AI maintain it… I’m sure a prompt like “Maintain this code base.” would do the trick… /s
All we then need is somebody to actually use the slop. Maybe AI could do that too. Seems best…
@samir But bottom line (almost literally) the whole thing seems to be about getting it funded…
“Perhaps the biggest barrier, however, will be convincing the U.S. government to fund the project, says Ji, particularly at the scale envisaged. She thinks a more realistic goal might be to solicit funding from the private sector for a proof of concept.”
It’s a get rich quick plan. 🤮
-
@mkoek It does, but I dispute that this is any less secure than e.g. GitHub.
@samir AI often hallucinates package names. Bad guys squat malware on those names. I don’t think that’s easily done with a C/C++ codebase. With AI-translated Rust code, it could happen. A bigger risk than potential buffer overflows in a mature C program, I think.
-
@samir AI often hallucinates package names. Bad guys squat malware on those names. I don’t think that’s easily done with a C/C++ codebase. With AI-translated Rust code, it could happen. A bigger risk than potential buffer overflows in a mature C program, I think.
@mkoek Yes, it’s true, but you could also generate a Cmake file which automatically downloads a malicious dependency from GitHub.
The problem is the “hallucination”, not the choice of programming language.
-
@samir But bottom line (almost literally) the whole thing seems to be about getting it funded…
“Perhaps the biggest barrier, however, will be convincing the U.S. government to fund the project, says Ji, particularly at the scale envisaged. She thinks a more realistic goal might be to solicit funding from the private sector for a proof of concept.”
It’s a get rich quick plan. 🤮
@fubaroque You get it.
-
RE: https://mastodon.social/@ieeespectrum/116030177478995194
What is the point of converting this software?
Who’s gonna maintain it?
Unless you have interest from the maintainers, you have a bunch of code that no one uses and no one maintains. Good job, folks.
I see nothing on these sites that explain how they hope to drum up maintenance, or even engage the maintainers of the original code.
Once again, the critical part of FOSS is ignored. It’s the people.
it's ok. AI can't reliably just convert any code, so it probably won't run at all. which, one could argue, is the only way to be sure it's secure (if no one can ever run it).
-
@samir this field is deranged
@samir I had a guest and I played the above without reading the details. I insist, this field is deranged
-
@samir "The Institute for Progress" sounds like another "effective altruism" wankfest.
@gpcureton @samir Right on the money.
Their website uses AI-generated images for post thumbnails and promotes AI, their open position for an Editor explicitly says that you should be comfortable working with LLMs, it reminded me of the accelerationist movement, but since you've also mentioned effective altruism, it turns out there's a subset named "effective accelerationism" which is used in the corporate world.
IFP partially matches e/acc, they push for accelerated technological progress through US-centric reforms, they're also pro-AI and pro-innovation but they're not (or haven't revealed to be AI maximalists), so it's not explicitly e/acc, just smells a bit like it.
To further that, three of the foundations funding IFP - Astera Foundation, Coefficient Giving and Heising-Simmons Foundation are EA donors or EA-aligned, and IFP may have received funding in the past from none other than one of the poster boys of EA, Sam Bankman-Fried, at least according to an EA forum page on IFP.
-
@samir I had a guest and I played the above without reading the details. I insist, this field is deranged
@RosaCtrl It hasn’t changed since earlier!
-
@gpcureton @samir Right on the money.
Their website uses AI-generated images for post thumbnails and promotes AI, their open position for an Editor explicitly says that you should be comfortable working with LLMs, it reminded me of the accelerationist movement, but since you've also mentioned effective altruism, it turns out there's a subset named "effective accelerationism" which is used in the corporate world.
IFP partially matches e/acc, they push for accelerated technological progress through US-centric reforms, they're also pro-AI and pro-innovation but they're not (or haven't revealed to be AI maximalists), so it's not explicitly e/acc, just smells a bit like it.
To further that, three of the foundations funding IFP - Astera Foundation, Coefficient Giving and Heising-Simmons Foundation are EA donors or EA-aligned, and IFP may have received funding in the past from none other than one of the poster boys of EA, Sam Bankman-Fried, at least according to an EA forum page on IFP.
@gpcureton @samir But wait, it gets better (or worse?).
I checked IFP on ProPublica, in the 2022 and 2021 Form 990 filings (https://projects.propublica.org/nonprofits/organizations/872449729/202300699349300540/full), the org name is "INSTITUTE FOR PROGRESS AND ALTRUISM".
Part III 4a mentions "[...]. As a non-profit research and advocacy organization, IFP brings the best ideas from the progress studies and effective altruism communities to policymakers in Washington, DC."
In their 'preparing for launch' page they mention they borrow ideas from defensive accelerationism (https://ifp.org/preparing-for-launch/).
So yeah, IFP is that type of woo-woo techbro policy-wonk.
-
@gpcureton @samir But wait, it gets better (or worse?).
I checked IFP on ProPublica, in the 2022 and 2021 Form 990 filings (https://projects.propublica.org/nonprofits/organizations/872449729/202300699349300540/full), the org name is "INSTITUTE FOR PROGRESS AND ALTRUISM".
Part III 4a mentions "[...]. As a non-profit research and advocacy organization, IFP brings the best ideas from the progress studies and effective altruism communities to policymakers in Washington, DC."
In their 'preparing for launch' page they mention they borrow ideas from defensive accelerationism (https://ifp.org/preparing-for-launch/).
So yeah, IFP is that type of woo-woo techbro policy-wonk.
@alextecplayz @gpcureton Fucking hell.
I bet they make a lot of money from this bullshit, and it produces exactly zero useful pieces of software.
-
@samir "rust rewrite" is rapidly becoming a nebulously security-flavored term that people lacquer onto something indiscriminately, like "encryption" or "vpn". all things that are real and work and are good, but lacking any of the context of when or why they are good
-
@samir Ah, but the AI will maintain it.
That way, we can bring more FOSS under oligarch control!
-
RE: https://mastodon.social/@ieeespectrum/116030177478995194
What is the point of converting this software?
Who’s gonna maintain it?
Unless you have interest from the maintainers, you have a bunch of code that no one uses and no one maintains. Good job, folks.
I see nothing on these sites that explain how they hope to drum up maintenance, or even engage the maintainers of the original code.
Once again, the critical part of FOSS is ignored. It’s the people.
@samir @rogersm @ieeespectrum It’s also a pity that people seem to have forgotten about the foundations of computer science. The Halting Problem is still a thing…
-
RE: https://mastodon.social/@ieeespectrum/116030177478995194
What is the point of converting this software?
Who’s gonna maintain it?
Unless you have interest from the maintainers, you have a bunch of code that no one uses and no one maintains. Good job, folks.
I see nothing on these sites that explain how they hope to drum up maintenance, or even engage the maintainers of the original code.
Once again, the critical part of FOSS is ignored. It’s the people.
This can only end well...
-
@mkoek Yes, it’s true, but you could also generate a Cmake file which automatically downloads a malicious dependency from GitHub.
The problem is the “hallucination”, not the choice of programming language.
@samir agree that that would be equally bad, but have not seen people be that silly… that’s probably me though
-
@samir agree that that would be equally bad, but have not seen people be that silly… that’s probably me though
@samir Goodness, I am behind the times. Some searching tells me that this is the way people handle dependencies now - just pull in some github repo from your makefile. We are not getting better at this security thing, are we.
-
@samir Goodness, I am behind the times. Some searching tells me that this is the way people handle dependencies now - just pull in some github repo from your makefile. We are not getting better at this security thing, are we.
@mkoek Nope. It’s the same everywhere. And I can’t fault it; it can take between “months” and “never” for a dependency to arrive as a Debian package. (And sometimes you do need an exact version.)
I personally like nixpkgs as a solution to this, but it’s a big change for most people.
-
@mkoek Nope. It’s the same everywhere. And I can’t fault it; it can take between “months” and “never” for a dependency to arrive as a Debian package. (And sometimes you do need an exact version.)
I personally like nixpkgs as a solution to this, but it’s a big change for most people.
@samir True. Apparently that’s the price of having some checks in place.
