AI Wars

[SOL]

A new artificial intelligence model developed by Chinese researchers is performing untold feats with image creation and natural language processing — making rivals in Europe and the U.S. nervous about falling behind.

The model, dubbed Wu Dao 2.0, is able to understand everything people say — the grammar too — but can also recognize images and generate realistic pictures based on descriptions. It can also write essays and poems in traditional Chinese, as well as predict the 3D structures of proteins, POLITICO'S AI: Decoded reported.

Developed by the government-funded Beijing Academy of Artificial Intelligence and unveiled last week, Wu Dao 2.0 appears to be among the world's most sophisticated AI language models.

Wu Dao 2.0's creators say it's 10 times more powerful than its closest rival GPT-3, developed by the U.S. firm OpenAI.


Massive language models, which produce text that looks like it could have been written by a human — made famous in a Roald Dahl short story that was fiction at the time — are one of the most powerful AI-powered technologies.

Companies like Google are betting on such models as a way to revolutionize online search, opening the way for a future in which consumers could ask their devices anything, and obtain responses that seem to be written by an expert.

The language models are also an informal indicator of countries' advancement in AI research — a key aspect of tech competition between China, the United States and Europe.

Whoever develops the most potent language model will be best positioned to corner future markets and spread influence — not to mention the bragging rights.

AI models also reflect the data and biases of their programmers, meaning that English and Chinese language models are set to dominate — casting a shadow over other cultures that are fighting to keep their languages in wide use. Critics warn these language models could also be used as powerful disinformation and propaganda machines.


Ahead of a visit to Europe by U.S. President Joe Biden that is set to focus heavily on tech-related matters, EU policymakers and industry groups have sounded the alarm about China's growing sophistication in AI and Europe's relative lag.

“We are about to lose digital sovereignty in the AI space if we don’t act right now,” said Jörg Bienert, chairman of Germany’s Federal AI Association, an industry group.

If Europe doesn’t start developing large language models itself, he added, “we might see monopolies like we have with Google in search engines” because of the huge computing, data and human resources the models require.

German conservative MEP Axel Voss shared Biernet’s anxiety about language models, tweeting that China “beat the US to it, while the EU was not even close in the race…”

Show me the money
In April, AI industry associations from seven European countries — Germany, Austria, Sweden, Croatia, Slovenia, the Netherlands, France and Bulgaria — sent a letter, seen by POLITICO, to the EU's Internal Market Commissioner Thierry Breton asking for money to develop large AI models.

Some countries, such as France, have programs up and running. The industry group is about to launch an initiative called Large European AI Models, which will develop models in several European languages. The group is "gathering comrades for combat," Bienert said. But there aren't many programs like it.

Brave new world: The first 100 days of Liz Truss’ Britain
The EU’s executive arm has allocated money specifically to AI through various funding programs, such as the Horizon 2020 and its long-term budget.

But the European Commission itself acknowledges that more is needed. According to a study it published with the European Investment Bank, the U.S. and China cover 80 percent of investments in AI and blockchain technologies, while Europe invests 7 percent, which totals around €1.75 billion.

The report found that Europe has a €10 billion investment gap to the U.S. and China in artificial intelligence and blockchain technologies.

Breton, the EU commissioner, is focused on getting EU countries and firms to share more industrial data with each other — a pet project of his, which he argues will feed into even better artificial intelligence solutions. Breton said the EU is working toward “giving itself the means to control” the quality of data, computing power and algorithms needed to run AI systems.

The European Commission has also proposed rules to promote commercial and industrial uses of AI, while ensuring it doesn't harm people. The bloc is also unveiling new supercomputers this year.

Ex-Google CEO Eric Schmidt, who chairs an advisory committee meant to explore how AI could benefit U.S. national security, has called the EU’s approach a “disaster,” and accused the bloc of not doing enough to counter the “Chinese threat.” China beating the U.S. in AI tech is exactly the kind of scenario that he dreads, and which has prompted him to argue for more government spending on military AI applications by the U.S. and its allies.

Bienert said the West should act now. "Especially in respect to what we got to know from China, I think we need a collaboration between the U.S. and Europe in this area," he said.

https://www.politico.eu/article/meet-wu ... est-sweat/

[Budge]

A new artificial intelligence model developed by Chinese researchers is performing untold feats with image creation and natural language processing — making rivals in Europe and the U.S. nervous about falling behind.

The model, dubbed Wu Dao 2.0, is able to understand everything people say — the grammar too — but can also recognize images and generate realistic pictures based on descriptions. It can also write essays and poems in traditional Chinese, as well as predict the 3D structures of proteins, POLITICO'S AI: Decoded reported.

Developed by the government-funded Beijing Academy of Artificial Intelligence and unveiled last week, Wu Dao 2.0 appears to be among the world's most sophisticated AI language models.

Wu Dao 2.0's creators say it's 10 times more powerful than its closest rival GPT-3, developed by the U.S. firm OpenAI.


Massive language models, which produce text that looks like it could have been written by a human — made famous in a Roald Dahl short story that was fiction at the time — are one of the most powerful AI-powered technologies.

Companies like Google are betting on such models as a way to revolutionize online search, opening the way for a future in which consumers could ask their devices anything, and obtain responses that seem to be written by an expert.

The language models are also an informal indicator of countries' advancement in AI research — a key aspect of tech competition between China, the United States and Europe.

Whoever develops the most potent language model will be best positioned to corner future markets and spread influence — not to mention the bragging rights.

AI models also reflect the data and biases of their programmers, meaning that English and Chinese language models are set to dominate — casting a shadow over other cultures that are fighting to keep their languages in wide use. Critics warn these language models could also be used as powerful disinformation and propaganda machines.


Ahead of a visit to Europe by U.S. President Joe Biden that is set to focus heavily on tech-related matters, EU policymakers and industry groups have sounded the alarm about China's growing sophistication in AI and Europe's relative lag.

“We are about to lose digital sovereignty in the AI space if we don’t act right now,” said Jörg Bienert, chairman of Germany’s Federal AI Association, an industry group.

If Europe doesn’t start developing large language models itself, he added, “we might see monopolies like we have with Google in search engines” because of the huge computing, data and human resources the models require.

German conservative MEP Axel Voss shared Biernet’s anxiety about language models, tweeting that China “beat the US to it, while the EU was not even close in the race…”

Show me the money
In April, AI industry associations from seven European countries — Germany, Austria, Sweden, Croatia, Slovenia, the Netherlands, France and Bulgaria — sent a letter, seen by POLITICO, to the EU's Internal Market Commissioner Thierry Breton asking for money to develop large AI models.

Some countries, such as France, have programs up and running. The industry group is about to launch an initiative called Large European AI Models, which will develop models in several European languages. The group is "gathering comrades for combat," Bienert said. But there aren't many programs like it.

Brave new world: The first 100 days of Liz Truss’ Britain
The EU’s executive arm has allocated money specifically to AI through various funding programs, such as the Horizon 2020 and its long-term budget.

But the European Commission itself acknowledges that more is needed. According to a study it published with the European Investment Bank, the U.S. and China cover 80 percent of investments in AI and blockchain technologies, while Europe invests 7 percent, which totals around €1.75 billion.

The report found that Europe has a €10 billion investment gap to the U.S. and China in artificial intelligence and blockchain technologies.

Breton, the EU commissioner, is focused on getting EU countries and firms to share more industrial data with each other — a pet project of his, which he argues will feed into even better artificial intelligence solutions. Breton said the EU is working toward “giving itself the means to control” the quality of data, computing power and algorithms needed to run AI systems.

The European Commission has also proposed rules to promote commercial and industrial uses of AI, while ensuring it doesn't harm people. The bloc is also unveiling new supercomputers this year.

Ex-Google CEO Eric Schmidt, who chairs an advisory committee meant to explore how AI could benefit U.S. national security, has called the EU’s approach a “disaster,” and accused the bloc of not doing enough to counter the “Chinese threat.” China beating the U.S. in AI tech is exactly the kind of scenario that he dreads, and which has prompted him to argue for more government spending on military AI applications by the U.S. and its allies.

Bienert said the West should act now. "Especially in respect to what we got to know from China, I think we need a collaboration between the U.S. and Europe in this area," he said.

https://www.politico.eu/article/meet-wu ... est-sweat/

The problem of new tech is having geniuses trying to test it:

https://www.youtube.com/watch?v=0XTaOmNqlF0

[SOL]

The problem of new tech is having geniuses trying to test it:

That was hilarious, and boy was the old boy lost. He does not mince too many words does he?

Here is a follow up

https://www.youtube.com/watch?v=orD-e_W6Pic

[Budge]

The problem of new tech is having geniuses trying to test it:

That was hilarious, and boy was the old boy lost. He does not mince too many words does he?

Here is a follow up

https://www.youtube.com/watch?v=orD-e_W6Pic

Classic. Had me howling.

Used to have a laugh when US shows interviewed folks from Belfast, Glasgow, Liverpool etc. and had to use subtitles.

"Indeed, in many respects, she was quite English, and was an excellent example of the fact that we have really everything in common with America nowadays, except, of course, language." Oscar Wilde's “The Canterville Ghost”

[harryg]

I don't know what was the funniest part of that. Her "English" accent was a cracker.

On another note, there was a time when people were worried about being spied on, now they actually pay for a little electronic spy to come into their house and listen to them...

It's not quite "1984", but getting there.

[Budge]

I don't know what was the funniest part of that. Her "English" accent was a cracker.

On another note, there was a time when people were worried about being spied on, now they actually pay for a little electronic spy to come into their house and listen to them...

It's not quite "1984", but getting there.

No Harry, I think we're already at 1984.

[MarkD]

For all the youngsters out there, we were likely past 1984 by 1984.

[Cinnamon]

It is worth mentioning that the EUV lithography machine developed by Russia may be more advanced than ASML's EUV lithography machine.

We know that the most advanced lithography machine at present is EUV, and only ASML can produce it. Because ASML's EUV technology mainly comes from the EUV alliance led by the US company Intel, EUV shipments are limited.

EUV lithography machines cannot be sold to Chinese companies, let alone Russian companies. Although Russian companies can't use them now, they have already laid out the research and development of EUV lithography machines. So, is there any advantage for Russia to study EUV, and is it possible to succeed?

First, Russia has technological advantages. ASML's most advanced EUV lithography machine uses EUV extreme ultraviolet light source with a wavelength of 13.5nm, which is mainly used to manufacture chip production lines with advanced process technology of 7nm and below.

The new EUV lithography machine being developed by Russia adopts another design route and will use X-ray technology. Its wavelength is between 0.01nm and 10nm, which is shorter than EUV extreme ultraviolet light, and the lithography resolution higher rate.

In addition, the chip can be produced without the need for a photomask. According to local media, these two points are stronger than ASML.


Second, Russia has the talent advantage. This time, the Russian EUV lithography machine undertakes the research and development of the Russian national research universities and colleges. It is a leader in the fields of microelectronics and nanoelectronics. There are many R&D talents in this field, and their strength is still very strong.

Although the chip industry in Russia is underdeveloped, their scientific and technological capabilities are not bad, because they have the world's top and largest number of math talents, which is unmatched by other places. Huawei's 3G technology breakthrough is the math brother in Russia.

Now, Russia has invested a lot of money for this. If it is devoted to research and development, it is really possible to break through the EUV lithography machine

Furthermore, Russia has the advantage of cooperation. Today, EUV lithography machines have been used as a tool for sanctions. If Russia really wants to break through the EUV lithography machine, I believe we will actively cooperate to break the Western monopoly on advanced lithography machines.

Although Russia has the above three advantages, it is not easy to surpass it. According to industry insiders, the X-ray lithography machine still has certain limitations. The production efficiency cannot be compared with ASML's EUV lithography machine, and it is only suitable for specific scenarios.

In addition, it is said that due to the recent situation, as many as 70,000 experts from Russia have left, which will naturally affect subsequent research and development.

https://min.news/en/world/f4ef001773ca3 ... ec7f9.html

What do you guys think of this story and the others that will follow; any truth to it

[SOL]

Without China's help or barring some massive new development like finding a way to chain slower chips together (kind of like parallel computer processing) to beat smaller chips, it seems that this is wishful thinking for now. Tom's hardware confirms this. Russia is seeking to have 28nm by 2030. That is essentially a stone age type of thinking in today's timelines.

As it's being ostracized and sanctioned by much of the world for its war against Ukraine, Russia is building up plans to revive its ailing local manufacturing of semiconductors since it cannot get chips from the usual suppliers. The country's new chip plan involves a rather massive investment over the next eight years. The goals don't exactly sound ambitious. For example, while TSMC plans to hit 2nm by 2026, Russia wants 28nm local chip manufacturing by 2030.

Russia's government has developed a preliminary version of its new microelectronics development plan that requires investments of around ₽3.19 trillion ($38.43 billion) by 2030. The money will be spent on the development of local semiconductor production technologies, domestic chip development, datacenter infrastructure, developing of local talents, and marketing of homebrew chips and solutions, reports Cnews.

On the semiconductor manufacturing side of matters, the country plans to spend ₽420 billion ($5 billion) on new fabrication technologies and their ramp-up. One of the short-term goals is to ramp up local chip production using a 90nm fabrication technology by the end of the year. A longer-term goal is to establish manufacturing using a 28nm node by 2030, something TSMC did in 2011.

Being historically rather successful with software and high-tech services, Russia has been comparatively unsuccessful with chip design and manufacturing. While there are plans to educate local talent and develop chips domestically, one of the things that the country plans to do by the end of the year is to establish a reverse engineering program of 'foreign solutions' to transfer their manufacturing to Russia. All digital items should be produced domestically by 2024. Things that the country cannot make domestically are expected to be sourced from China.

https://www.tomshardware.com/news/russi ... -plan-28nm

[MarkD]

Those of us old enough to remember recall the Soviet space program and it's archaic systems, like rope and pull mechanisms.

They worked but.....call it what you like. One reason why they invest so much in human capital to educate in mathematics and science. US kids turn on YT or another app.

[outof thebox]

Those of us old enough to remember recall the Soviet space program and it's archaic systems, like rope and pull mechanisms.

They worked but.....call it what you like. One reason why they invest so much in human capital to educate in mathematics and science. US kids turn on YT or another app.

Math and science can help problems that seem unsolvable especially if you are focusing on humanities and arts. Nothing wrong with that but 90% of college grads (my opinion) pay a fortune to get degrees that are essentially expensive toilet paper.

Who knows maybe the Ruskies will pull a Houdini

[harryg]

Those of us old enough to remember recall the Soviet space program and it's archaic systems, like rope and pull mechanisms.

They worked but.....call it what you like. One reason why they invest so much in human capital to educate in mathematics and science. US kids turn on YT or another app.

I heard this one, don't know if it is a joke or true, don't want to find out because it would spoil the fun.

NASA supposedly spent $millions as part of their space programme developing a special pen that would work in zero gravity.

The Russians used a pencil.

I like this kind of joke/truth. It shows or suggests that common sense can very often outperform academic brainitude.

[jlhooter]

Harry I need to hang with you one day, buy drinks and just laugh the night away.

[Triplethought]

Those of us old enough to remember recall the Soviet space program and it's archaic systems, like rope and pull mechanisms.

They worked but.....call it what you like. One reason why they invest so much in human capital to educate in mathematics and science. US kids turn on YT or another app.

I heard this one, don't know if it is a joke or true, don't want to find out because it would spoil the fun.

NASA supposedly spent $millions as part of their space programme developing a special pen that would work in zero gravity.

The Russians used a pencil.

I like this kind of joke/truth. It shows or suggests that common sense can very often outperform academic brainitude.

Here is the background on that. https://www.scientificamerican.com/arti ... en-able-t/

[Yodean]

Here is the background on that. https://www.scientificamerican.com/arti ... en-able-t/

But I thought the Earth was flat ... why would you need a pen in space?!