jameshilliard

> If you only accept papers that were published in reputable venues and analyze LLMs that are less than a month old, you are left with zero useful papers (since writing a useful paper takes longer than a month).

Why would one look only at formal published scientific papers? Especially if the only publications are known to be quite outdated and highly likely to be no longer particularly relevant anymore.

> It is impossible to have an intellectual discussion if all scientific material is disregarded.

One can certainly have an intellectual discussion even for areas without relevant formal published scientific papers.

Certainly one can do so if the subject matter is in the area of your own expertise, that can even lead to scientific papers being written.

I'm an experienced open source developer leveraging AI tools for linux kernel, u-boot and buildroot contributions in addition to reverse engineering, so clearly this subject is well within my area of expertise regardless of what some outdated published research says.

I probably wouldn't even disagree with that paper's assessment at the time, however the situation is very different now. I have noticed that there is an issue with many experienced open source developers simply having not figured out how to properly leverage AI for open source projects they work on. Probably one reason for this is that productive workflows with AI are often very different from previous workflows and it takes some practice to figure out how to best use AI tools, these best practice can also change over time.

It's likely significant percentages of the debian userspace is written using some form of AI assistance as well.

> how are they different? did the algorithms upon which they operate suddenly change?

Just like most technologies AI algorithms and tools get continuously refined/improved over time.

One thing modern AI has gotten especially good at in the last year is reverse engineering, it went from struggling and frequently making mistakes when analyzing binaries even with say a specialized Ghidra MCP integration to being able to very quickly and accurately analyze even heavily stripped BROM binaries with almost no user guidance or specialized environments. Binaries that would previously take days or weeks to analyze can be analyzed by a modern AI in minutes to hours.

The difference is massive and obvious to anyone who has been working with these tools over these time periods.

> No. No it is not, despite how many times AI proponents like to claim it is.

Why do you think there isn't a huge difference? I've used AI tools over this time period and the difference is very obvious.

> For one it's a 78 year old colony that excludes the indigenous Palestinian population from returning to their ethnically cleansed homes.

The ethnic cleansing that happened when Israel gained independence was not in just one direction, Israel also did allow plenty of Arab/Palestinians to stay and gave them citizenship where feasible. It's not surprising however that Israel doesn't currently offer citizenship to a population that largely doesn't think Israel should even exist. States don't normally give citizenship to their non-citizen enemies.

> They also allow people to convert to Judiasm and enter, so it actually has nothing to do with their ethnicity and is solely based on religion.

It's not solely based on religion. It's also quite rare for people to convert to Judaism in general so that exception has little practical implications.

> Israel refuses to even define its borders. And current behaviour demonstrates why.

States like Israel formally define their borders via mechanisms like peace deals that include bilateral recognition of agreed upon borders, like Israel has with Egypt and Jordan.

You can't realistically define borders properly otherwise, trying to define borders via unilateral declaration doesn't really work when there are 2 states next to each other, the borders ultimately need to be negotiated bilaterally between those states with both sides recognizing the same borders.

> The other 94% was ethnic cleansing.

That's not really an accurate characterization of the situation as much of the land was uninhabited/unregistered and/or state owned. The Negev desert for example is a large percentage of the total land area and was effectively uninhabited and not privately owned. Land ownership records at the time were also not particularly clear and often there were disputed claims. Also Israel has about 20% Arab/Palestinian citizens so it's not like they ethnically cleansed everyone like Egypt/Jordan did with the Jews in Gaza and the West Bank when Israel gained independence.

> U-Boot upstream itself is staunchly anti-AI, at least at the time of this release, and much discussion of the above was had on their mailing list.

I saw this as well, which is funny because I've sent plenty of patches written using AI assistance to upstream U-Boot. It was my understanding that U-Boot was mostly following kernel conventions in general as many U-Boot drivers share code(in addition to the actual subsystem maintainers themselves) with their kernel equivalents.

> Copyright law applies to all citizens whether or not they've agreed to a contract.

Funny enough SpaceX just ignores copyright law a lot of the time by refusing to comply with some of the GPL licenses for their dishy firmware. I think Tesla does as well to some degree. They generally publish some code but not the complete corresponding source code for the firmware they actually ship.

> They just re-host it on chinese clouds.

Isn't it difficult to get access to Chinese clouds without having legal entities registered in China?