The US has announced new rules for exporting computer chips as Joe Biden concludes his presidency. These rules aim to prevent advanced technology from being sent to rivals like China.
Category
🗞
NewsTranscript
00:00A whole new world of art, created almost entirely by computer.
00:06We can use generative AI to create things like text, images or even music.
00:11Leading the way on the artificial intelligence front are the US and China,
00:16where even tea harvesting is becoming automated on a trial basis.
00:20As for Europe, it has some serious catching up to do
00:23with what is probably the most important technology of the future.
00:28French startup Nuke AI develops new solutions for improving efficiency at logistics companies.
00:36The difference between our AI and other providers is that we involve people.
00:42We don't need huge amounts of data to be fed into machines.
00:46Our focus is on human beings working together with machines.
00:51We use machines to work better.
00:57France wants to become a leader in AI and has been actively attracting investors.
01:02The head of Microsoft wants to invest 4 billion euros in a new data center in France,
01:07which will also provide a boost for a number of new smaller firms.
01:13Startups in France will benefit from the financial support and advice on offer.
01:21France is still some way off the industry's leaders
01:24when it comes to patent applications and investments in AI data centers.
01:30The French government should focus on what we're good at,
01:33instead of trying to compete in areas where the US and China have a head start.
01:40It should support AI technologies that take a different approach to those that are currently on vogue.
01:48Berlin, Germany's political and AI capital.
01:53A range of cutting-edge companies have set up operations in the city.
01:57Among the complex tasks being worked on here at JetBrains, for example,
02:01instead of reading a 50-page document, you can ask questions and get answers about specific points.
02:07Part of the problem is we approach these models like we're speaking to another person that can understand us.
02:12But really they're just sophisticated algorithms that need specific instructions.
02:16So changes in, say, the wording or the tone can give us drastically different outputs.
02:22For good results, the AI might need a prompt comprising several pages of text.
02:27Prompt engineering is a fairly new business with high demand and relatively few machine learning experts.
02:34And when it comes to salaries, too, Europe is struggling to compete with the US.
02:41In the United States, salary sites indicate that this job is on average earning about $130,000 a year.
02:47But many people work with prompting, most notably ML engineers.
02:51And this job requires a deep knowledge of ML models beyond just prompting.
02:55So ML engineers are earning on average about $175,000.
03:00But competition for the best human minds is not everything.
03:04The focus has really been different.
03:06So in the United States, the top companies that are producing the top models are there.
03:10But they're really releasing these models in a proprietary fashion, so you have to pay to use them.
03:14In China and in Europe, by contrast, a lot of the focus has been on open-source models.
03:19So releasing models that are free and anyone can use them.
03:22So interestingly, some of these open-source models are also becoming competitive with those released by the top US AI companies.
03:31The US is still clearly ahead in AI.
03:34But some of the multi-billion dollar investments there have failed to pay off.
03:42After a sluggish start, European companies are now starting to make up ground.
03:48Meanwhile, China is itself hoping to surpass the US.
03:53The race to outperform the competition for AI supremacy is not over.