Become a fan of Slashdot on Facebook

 



Forgot your password?
typodupeerror
×
AI

Meet Aleph Alpha, Europe's Answer To OpenAI (wired.com) 42

An anonymous reader quotes a report from Wired: Europe wants its own Open AI. The bloc's politicians are sick of regulating American tech giants from afar. They want Europe to build its own generative AI, which is why so many people are rooting for Jonas Andrulis, an easy-going German with a carefully pruned goatee. Ask people within Europe's tech bubble which AI companies they're excited about and the names that come up most are Mistral, a French startup that has raised $100 million without releasing any products, and the company Andrulis founded, Aleph Alpha, which sells generative AI as a service to companies and governments and already has thousands of paying customers. [...] Now 41, Andrulis spent three years working on AI at Apple before leaving in 2019 to explore the technology's potential outside the constraints of a big corporation. He set up Aleph Alpha in Heidelberg, a city in southwestern Germany. The company set to work building large language models, a type of AI that identifies patterns in human language in order to generate its own text or analyze huge numbers of documents. Two years later, Aleph Alpha raised $27 million, an amount that's expected to be dwarfed by a new funding round Andriulis hints could be announced in the coming weeks.

Right now, the company's clients -- which range from banks to government agencies -- are using Aleph Alpha's LLM to write new financial reports, concisely summarize hundreds of pages, and build chatbots that are experts in how a certain company works. "I think a good rule of thumb is whatever you could teach an intern, our technology can do," Andrulis says. The challenge, he says, is making the AI customizable so businesses using it feel in control and have a say in how it works. "If you're a large international bank and you want to have a chatbot that is very insulting and sarcastic, I think you should have every right." But Andrulis considers LLMs just a stepping stone. "What we are building is artificial general intelligence," he says. AGI, as it's known, is widely seen as the ultimate aim of generative AI companies -- an artificial, humanlike intelligence that can be applied to a wide range of tasks.

The interest Aleph Alpha has received so far -- the company claims 10,000 customers across both business and government -- shows it is able to compete, or at least coexist, with the emerging giants of the field, says Jorg Bienert, who is CEO of the German AI Association, an industry group. "This demand definitely shows it really makes sense to develop and provide these types of models in Germany," he says. "Especially when it comes to governmental institutions that clearly want to have a solution that is developed and hosted in Europe." Last year, Aleph Alpha opened its first data center in Berlin so it could better cater to highly regulated industries, such as government or security clients, that want to ensure their sensitive data is hosted in Germany. The concern about sending private data overseas is just one reason it's important to develop European AI, says Bienert. But another, he says, is that it's important to make sure European languages are not excluded from AI developments. [...]

This discussion has been archived. No new comments can be posted.

Meet Aleph Alpha, Europe's Answer To OpenAI

Comments Filter:
  • LOL

    • They have customers in sensitive industries who don't want the risk of information leaks to foreign places.

      Just down my street a local company makes a voice assistant; I know another local company selling a customer assistance chatbot. What do you find laughable about launching a tech startup? There is space for more than one player.

      How is your black Ford model T?

  • by Whateverthisis ( 7004192 ) on Wednesday August 30, 2023 @08:26AM (#63808324)
    So not really a LLM technical expert, or even a hobbyist; I just watch this from afar like everyone else. So honest question:

    Granted many in Europe speak English, but the concept of a "European" OpenAI plus LLM highlights a curious thing: how much does the actual language of an LLM actually come into play? There are vast differences between German, French, Spanish, Portugese, Magyar, Polish, Danish, Dutch, Norwegian, Swedish, etc. etc. Does the actual language that an LLM is built on matter in the construction of an LLM-based AI? Or is this not relevant at all?

    In theory if it does matter, one would think that a "European" OpenAI would be much harder to build, but also more valuable when done as the ability to translate concepts from one language to another can be quite important. Anyone who's ever had to file a patent in another country and describe your invention in a language that doesn't have good words or concepts for the science behind an invention knows what I'm talking about.

    • Current LLM's are just based on statistics how often a word follows another, so there is no concept or culture behind it. I think the main motivation is to have state of the art tech without sensitive data flowing to USA, (remember the wiretapping of Merkel).
    • Abstractly, no, LLM's are doing pattern matching and predicting the next word from the previous words based on a corpus of data.

      The details are in the forward and backwards look-ahead windows and there must be some per-language tuning but human languages stem from the same hardware everywhere so languages basically work the same way; there's even a huge debate of whether a single Amazon language violates a Chomsky rule. ML works out the details as patterns. Facebook's LLM can decompose and compose across l

      • If you wanna relate LLMs to how human language processing works, the more appropriate theoretical lens would be cognitive linguistics. Adele Goldberg has been drawing some interesting parallels between LLMs & human language processing, e.g. both work on probabilistic models rather than Chomskian UG deterministic models.
    • by AmiMoJo ( 196126 )

      It could be an advantage to develop a multi-lingual AI. As well as being useful for translation tasks, there are some unique challenges that English doesn't have, such as dealing with gendered language. German, for example, has masculine, feminine, and neutral words, as I understand it.

      Solving those problems could be valuable. Japanese has a similar issue with politeness, which domestic AI companies are trying to solve.

      • LLMs deal with morpho-syntactic agreement in English so why couldn't they do so in other languages. If fact, they already do. ChatGPT works well in several languages that are highly inflected in gender, number, & case, e.g. Spanish.
    • by gweihir ( 88907 )

      No idea. But having revised a DeepL translation (German to English) just recently, I guess you can not just slap a translation layer on top as that would create far too many additional problems and make the results even more stupid.

    • ChatGPT is already extremely multilingual. I regularly switch languages when using it, depending on what I'm doing or what I need to know. It even "speaks" dying languages, though perhaps less fluently due to the lack of material.

      What I haven't tested, is whether it automatically shifts material into the language you are using. For example, if I have a programming question, I always ask it in English, because the English knowledge base is bigger. If there is some fact available only in English source mater

      • by MobyDisk ( 75490 )

        ChatGPT is the best language learning tool I have ever used.

        First: you can just chat with it any time, anywhere, on any topic. Ex: Suppose you want to learn German, but none of your German-speaking friends like to talk about video games? Maybe you are about to order from a restaurant, and you want it to be a virtual waiter. ChatGPT will talk about whatever you want.

        Second: It can explain translations better than people can. You can ask it to translate something, and EXPLAIN the translation. So if it us

    • by Tom ( 822 )

      You are aware that ChatGPT will speak pretty much any language you throw at it?

      Aside from a few nearly dead indigenous languages, there is more than enough material available in any language to train an LLM. The three official languages of the EU are English, French and German. Everyone learns at least one of these in school (a different one if your native language is one of them). All official EU documents are published in at least these languages.

      the ability to translate concepts from one language to another

      That isn't how LLMs work. They don't have "concepts" of any

  • Wow, I love these first three sentences that frame the story. Europe needs its own LLM company *because* European regulators would rather regulate Europeans than Americans. Would this LLM company provided value with a product that people want to use? Who cares! EU regulators want a punching bag, and they’re willing to pay for it!
    • EU regulators want a punching bag, and theyâ(TM)re willing to pay for it!

      No, they're willing to force Europeans to pay for it. If these politicians were putting their own, personal money into it, things would be different. But "the people" can pay for it just fine.

      Doesn't mean anything will come of it, but they DID SOMETHING!

      (and they can blame the Americans for screwing it up)

      • How would EU citizens end up paying for regulations that an AI company be required to follow?

        • Seriously? You're kind of slow dude. It's because they paid to build the AI company that they then regulate.

          But this wouldn't be anything new, for a long time now Europe has been trying to astroturf tech companies that they sink millions into that ultimately go nowhere. Unfortunately for them, most Europeans aren't talented at anything, and the ones that are usually end up emigrating to the US.

          • Seriously? You're kind of slow dude. It's because they paid to build the AI company that they then regulate.

            Where does it say the AI company is funded by citizens?

        • How would EU citizens end up paying for regulations that an AI company be required to follow?

          Following regulations has a cost, and this "alpeh alpha" thingy is going to be propped up by taxpayer's money in a futile effort to remain at least somewhat competitive with American technology, so... who again is going to pay that cost? Yeah, the EU taxpayer.

    • Oh no those terrible draconian EU regulators with their workers protections and making companies honor warranties.

      • by WDot ( 1286728 )
        No problem with worker protections and companies honoring warranties. AI regulation is definitely deep into the “reaching diminishing returns” end of regulatory policy, as AI regulators are primarily concerned that AI never does anything offensive or performs statistically differently for different demographics, which means spending more time and money writing paperwork proving safety than...actually making the product itself. Maybe that should be the case for a pharmaceutical product, but why s
  • And artificial stupidity! Domestic crap is always the best! Take that!

  • "Aleph Alpha" - wouldn't that translate to "A A" in English? (Aleph being A in Hebrew; Alpha being A in Greek). Wouldn't Aleph Iota be more fitting?

    • by Xteve ( 4172073 )
      No, because Iota is part of the word Idiota, which exists in Spanish, Portuguese, Italian, and in very similiar forms in other languages...
    • Wrong languages. Take AI and read it backwards as you might in Arabic, and you see the letters aleph (arabic) and alpha (greek).

  • "The bloc's politicians are sick of regulating American tech giants from afar. "
    If I understand this correctly, what they want is some successful LOCAL businesses that they can rapaciously pillage for tax revenue and crush under their legislation?

    I mean it's a funny motivation to try to encourage entrepreneurship but good luck.

  • ...Aleph, before answering questions will organize a roundtable with other AI chatbots, consult national AI bots, issue a 30 pages long recommendation about how the user should formulate questions, and file a request for an AI budget increase.

GREAT MOMENTS IN HISTORY (#7): April 2, 1751 Issac Newton becomes discouraged when he falls up a flight of stairs.

Working...