EU launches 4 testbeds to put AI tech through its paces before it goes to market
- Reference: 1687947693
- News link: https://www.theregister.co.uk/2023/06/28/eu_launches_it_four_facilities/
- Source link:
Under the European Commission's Digital Europe Programme, the testing and experimental facilities (or TEFs for short) will investigate the risks and impacts of new AI technologies before they hit the market.
Machine learning algorithms, robots, or self-driving cars can thus be tested in simulation and in the physical world by researchers focused on four different areas: manufacturing, healthcare, agriculture and food, as well as cities and communities.
[1]
"Providers of AI-driven solutions get the opportunity to test their products in real environments to assess if they meet the customer needs," Valentina Ivanova, project coordinator of AI-Matters, the manufacturing TEF, [2]said in a statement. "By offering access to testing and experimentation infrastructure across Europe, we aim to accelerate the uptake of these solutions in the market."
[3]
The European Commission pledged €220 million ($241 million) to fund the TEFs for five years. Member states approved the idea to support developers as new rules and regulations come into effect. Earlier this month, the European Parliament passed the AI Act regulating applications depending on their risk level. The most harmful products or services carrying unacceptable risks are banned and cannot be deployed within the EU.
[4]OpenAI calls for tough regulation of AI while quietly seeking less of it
[5]Euro Parliament green lights its AI safety, privacy law
[6]Euro Parliament green lights its AI safety, privacy law
[7]Europe to vote on AI laws with potential 7% revenue fines
TEFs support the new legislation, which called for regulatory sandboxes. Through rigorous testing, developers can ensure their technology is trustworthy and complies with laws so it can be deployed and made commercially available. Lawmakers are also interested in the results as they shape future AI policies.
The AI-Matters TEF is looking at how machine learning software and robotics can make the manufacturing industry more efficient. TEF-Health is interested in developing healthcare products that comply with the EU's laws on storing and using patient data. Meanwhile, the agrifoodTEF is focused on improving food production with robotic tractors and more. Finally, the CitCom.ai TEF is looking at how AI can improve power and mobility in cities with self-driving cars or installing better telecommunications software.
The four testing facilities are expected to be open by January 2024, whilst some projects will start as early as next month. ®
Get our [8]Tech Resources
[1] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=2&c=2ZJxZHoD4kp@FpBt45W3PDwAAAMg&t=ct%3Dns%26unitnum%3D2%26raptor%3Dcondor%26pos%3Dtop%26test%3D0
[2] https://www.compute.dtu.dk/english/news/nyhed?id=3ed202b0-54b1-43cd-bab0-5d68c435e1b7
[3] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44ZJxZHoD4kp@FpBt45W3PDwAAAMg&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[4] https://www.theregister.com/2023/06/21/openai_government_regulation/
[5] https://www.theregister.com/2023/06/15/european_parliament_ai_act/
[6] https://www.theregister.com/2023/06/15/european_parliament_ai_act/
[7] https://www.theregister.com/2023/06/12/europe_to_vote_on_ai/
[8] https://whitepapers.theregister.com/
Well, not sure how it's all supposed to work, but certainly much better to have AI agents that are going to be unleashed to the public being tested by some sort of independent entity, rather than just rolled out by the developer/manufacturer with a "trust us, it will be fine".
If the tests are known, standardised and unchanging then probably yes, the AI will be biased to pass the tests. But then again if the tests are configured to real-world scenarios, getting the AI to behave 'properly' in these scenarios is a win... just as long as there is no 'test mode' ie the AI should not know its being tested
Re: a NEUKlearer HyperRadioProACTive IT START for AI
The European Commission's Digital Europe Programme surely must realise that any and all truly disruptive and revolutionary AI programmers/developers will probably always market their wares whenever extremely lucrative for them, and unbelievably valuable to others, even should investigative testing and experimental facilities [TEFs] pronounce them dangerous and a viable existential threat to status quo orders/established SCADA Systems Administrations ...... unless and until, of course, there be arrangements made to reflect and cover their free open market price valuation losses as compensation for the voluntary withdrawal, curtailment of further refined product development and future free market un availability of their wares.
Failure to accept those simple terms even whenever they may be realised as expensive, can so easily be as chicken feed compared to the very likely horrendous costs and almighty destruction incurred and wrought with the introduction and enthusiastic exercising of a dangerous and viable existential threat to status quo orders/established SCADA Systems Administrations.
Has realistic EU provision for such been agreed and arranged? And if not, why not, whenever all can so easily be lost to that which they have no command and control over ‽ .
As well and good as this is, all that will happen is that they will bias the AI to pass the tests. Kind of like the Volkswagen emissions testing cheat.