OpenEuroLLM, an 'open, compliant, diverse' series of foundation models
slator.com
slator.com
Like these projects, OpenEuroLLM will go nowhere.
It is always the same approach that never works: First they assemble the usual mix of organisations from all sorts of countries, making sure all the major EU member nations are represented. Then they give the project some moronic goals (they state “compliance” as the #1 objective - wtf?).
The EU commission will pour millions in taxpayer money into this, which is nevertheless not remotely enough to match with the funds raised by US AI companies. And half the money for OpenEuroLLM will be spent on extensive bureaucracy where they have to write lengthy progress reports, pass audits by EU bureaucrats, etc. Not to mention the sizeable travel budget for the people involved. And finally, they may end up with prototypes that (1) nobody asked for and for which there is no market, (2) which do not work well/at all and (3) which are abandoned once the project ends.
By contrast, what the EU should absolutely, categorically NOT do is to create projects like this one. With an absurd number of participating organisations, dysfunctional project goals (like “compliance”) and the ballooning bureaucracy that ensues. What talent are they hoping to attract to this project? Any AI researcher with half a brain would avoid OpenEuroLLM, lest he (or she) drag down his (or her) career and waste precious lifetime for what is guaranteed to turn into another tombstone on Europe’s graveyard of ridiculous publicly funded technology projects.
The cookie law made me think that most people in the EU who make these laws need help turning on their computer.
I actually agree with parts of the Digital Markets Act, I should have the freedom to install whatever software I want on my iPad which is crippled by the App Store limitation, it makes my blood boil when I think about the fact that my iPad has a better CPU then my developer laptop, yet I can't run anything on it, because nanny Apple doesn't want me to. We'll see if it will work out though, I don't have high hopes.
I always say that the EU should look at parts of the US that are successful with an open mind and try to understand what's missing, because we have the talent that's not the issue, there is also money in Europe but for some reason I don't feel like I'm going to be building the next Google here. London got pretty close, but the UK decided to Brexit.
I agree that the EU should not try to create innovation, they should just create the conditions necessary for it.
The US also has bureaucracy issues, but it seems that there is so much investor money out there that you can just throw it at the problems to make them go away.
On the flip side, consider the economic cost that was caused by the GDPR. There are compliance and reporting obligations that require time and personell. There are additional development and monitoring costs. There is an impact on business models that have made it more difficult to monetise a user base. And lastly, there is a huge legal risk. Courts are scrambling to interpret the GDPR and the EU commission is happy to dish out fines (albeit mostly against American Big Tech companies).
On the bottom line, even well-intended legislation like the GDPR may have huge economic costs with little tangible benefits. And that’s not to speak about later legislation like the DSA (censorship) law, the DMA, the AI Act. And worst of all, the supply chain law (even though that’s unlikely to be as much of an issue for the software industry).
As for EU research projects, I agree with most that has been said, but there are some very good exceptions. The problem with EU research programmes is that they offer often too research and little businesses opportunities.