Chapter 523 Tower of Babel (4827)

Breaking the Language Matrix! ?

Sergey Brin shuddered and immediately returned his attention to the densely packed graphs on the screen.

Snatching the mouse from Demis Hassabis, Sergey Brin kept sliding the mouse wheel down.

Da Zhou language, Prussian language, Gaulish language, English language, Japanese language, Mao language, Xibai language, Portuguese language, Italian language, Barat language...

More than 300 curve comparison pictures all illustrate this fact:

The performance curves of Juzi2.5 in fifteen languages ​​are almost all on the same level!

Whether it is understanding and memory, reasoning and cognition, autonomous planning and decision-making, self-optimization and learning, emotion and social simulation, tool calling...

Except for the large fluctuations in generation and expression, the performance of almost all other abilities, especially reasoning and cognition, is almost exactly the same in various language environments!

There is not even a one percent difference!

How is this possible!?

This completely goes against the principles of the big model!

"Did they adjust the parameters to bring the performance of all languages ​​into line? Or did they translate into English first, think in English, and then translate back into the thought chain?"

Sergey Brin felt sweat suddenly appear on his head and under his armpits.

Artificial intelligence, the so-called intelligence, is thinking, and thinking requires language. Whether it is humans or computers, it is impossible to think without language.

This is especially true for large-model AI.

These large models may master all the languages ​​in the world, but when thinking in different languages, the performance of the large models in different aspects is different.

Firstly, it is because of the differences in the amount and quality of training materials for different languages.

In today's Internet age, Anglophone corpus is of course the most abundant, accounting for more than 80% of the total amount of data on the Internet.

When AlphaZero uses English for understanding and reasoning, its accuracy rate is more than 5% higher than that of other languages.

Secondly, different languages ​​themselves have different "expression preferences" and "performance differences".

For example, Prussian is faster than English in structural reasoning, while Spanish has a clear advantage in sensory corpus.

Large models usually use a single language to build their reasoning path in a single thinking loop.

Although it can recognize multiple languages ​​at the input stage and translate at the output stage, its inherent cognitive tensor structure still tends to use the token space constructed by the input language for semantic calculations.

To put it in simple terms, when the big model thinks about a problem, in a thinking loop, it basically only uses one language to think. If you use English, it will think in English. If you use Zhouwen, it will think in Zhouwen.

Even if it mixes other languages ​​into its replies, it is just a citation of materials or an imitation of human writing style, not true cross-lingual thinking.

How is it possible that the Orange model has similar performance in all aspects when thinking in different languages?

This is totally against common sense!

The only possibility is to align the thinking performance of various languages.

To put it simply, it is like a wooden barrel. Take the shortest board as the base and cut off all the tall boards.

But what’s the point of doing this?

Demis Hassabis shook his head hesitantly: "Probably not. Lacy's operation is a waste of resources."

"As for whether to translate the thoughts into English or another language first..."

Demis Hassabis paused.

"I thought so at first, but after testing, it turned out not to be the case."

After saying that, Hassabis turned to the middle of the lab report.

"Look at the 'abstract induction' and 'formal reasoning' capabilities. Even when using 'Malayalam', Juzi 2.5 can still accurately perform abstract induction and formal reasoning."

"For example, in this example, when it comes to emotion understanding, our experimenters asked Juzi to think in Malay and output the results in English. In its response to the experimenters' task, the Orange model did not simply translate the Malay word 'manja' into 'pampered' or 'affectionate.'"

"Instead, it uses different expressions with actual semantics."

"For example, in the first paragraph, it translates 'manja' as 'cute and clingy'."

"In the fifth paragraph, the thought chain also uses 'manja', but because the subject has changed, the meaning of 'manja' in Malayalam has also undergone a subtle change. Now, it transforms 'manja' here into 'being pampered'."

Demis Hassabis took off his glasses, wiped them, and squinted. "There's no exact translation for manja in Anglophone. But after this translation, even a Malayalam who has never learned Anglophone can speak Anglophone without any misunderstanding."

Sergey Brin looked at the graph that Demis Hassabis pointed to, and the hairs on his body stood up.

As a tech geek and one of the bosses of Google, he is certainly not the best in technology now, but his understanding and cognition of artificial intelligence are definitely among the best in the world.

How is this possible?

Because Malayalam is an isolating language, its grammatical structure is relatively flat, and its cultural context is more colloquial and situation-driven, which makes it inherently limited in its expression in abstract, philosophical, technical and other fields.

This results in the lack of some high-level conceptual vocabulary in the language itself, and often we can only rely on descriptive translation or the direct introduction of foreign words.

Philosophical imagery words like "consciousness", "existence", "subjectivity" and "objectivity" do not exist in Malayalam and can only be directly borrowed from foreign words or

But at the same time, there are also a considerable number of "flexible words" in Malayalam that are not found in Anglophone and Western languages.

The meanings of these words are often very subtle, and the corresponding words in the dictionary are more or less different.

This results in subtle differences in AI's understanding of the world and relationships when different corpora are used to train large models, or when large models are used in different languages.

This "subtle difference" may seem insignificant, but in fact it is often one of the important causes of cultural misunderstanding and conflict.

"Sergey, there's something more representative here, which is its understanding of 'sin'."

Demis Hassabis tapped the touchpad and swiped upwards. "Look here! This is a comparison of Zhou Wen and Ang Wen's thought chains on the same topic, as discussed by Juzi2.5G."

"Oh, maybe you don't know that sin, in Zhouwen, is generally translated as 'crime', but the meaning of the word 'crime' in Zhouwen is not the same as sin."

Demis Hassabis is a mixed-race Zhou person who knows a little bit of Zhou language. He originally couldn't distinguish the subtle semantic differences here.

But for a genius like him, once he realized the problem and started researching this area, it wouldn't take long for him to understand this conceptual difference clearly.

"But all along, translators have basically ignored this point and simply mechanically translated the Zhou version of '罪' as 'sin' and the Ang version of 'sin' as '罪'."

"Juzi 2.5 is different. When explaining legal issues, it translates sin as 'crime' in Zhouwen."

"In the later question of faith, it uses at least six different expressions of sin in Zhouwen, depending on the context: 'disobedience,' 'offense,' 'fault,' 'ungratefulness,' 'evil ways,' and 'suffering.'"

"These different expressions, in their respective contexts, accurately capture the most accurate meaning of the word 'sin' in that context. Subtle differences in word choice and sentence structure will rarely lead to incorrect subjective judgments on the other side."

"Oh, by the way, even the two common words 'subjective' and 'judgment' show subtle semantic shifts in Zhouwen and Angwen."

Demis Hassabis' eyes were filled with shock after careful consideration.

Sergey Brin frowned.

He certainly understood every term Demis Hassabis used.

But he has not yet established a complete causal mapping between "Hassabis's linguistic introduction" and "the convergence of Juzi2.5's thinking performance under multilingual conditions."

This just seems like... better translation software?

What does it have to do with the performance of the Orange large model?

No, this performance is beyond the capabilities of existing multilingual models—there must be some mechanism behind it that we haven't grasped.

This mechanism enables the Orange model to deeply understand the precise meaning of different languages ​​in different contexts, and even uses "explanation substitution" and "tone fitting" in translation to achieve more accurate translation of the original text.

Wait, what did Demi say before looking at this lab report?

Breaking through the language matrix?

I was a little confused before about what this language matrix was.

So...

"Demi, what you mean is... Juzi2.5's thinking is not based on a specific language, but... but..."

Sergei was so anxious that he scratched his head and grimaced, but he just couldn't find a suitable word to describe his understanding.

"It is a method that completely breaks down language boundaries and uses all the languages ​​in the world to form a 'high-semantic mapping map'."

Demis Hassabis took a deep breath and added to Sergei.

"High semantic mapping atlas! That's right! This is it!"

Sergey Brin slapped his thigh hard!

"However, this term is still too technical. To put it more figuratively... it combines languages ​​from all over the world to create an unambiguous language that only it can use and understand!"

"A language that transcends language families and semantic divisions... this kind of language can truly be called a 'whole language'."

After saying this, Sergey Brin's face turned pale. He wiped the sweat from his head, his eyes struggling, as if he still refused to believe it.

"Omni-language, OMG, does such a language really exist?"

"Even in a language only AI can use?"

"But the problem is that Juzi1.99DEC is open source and doesn't have this functionality at all."

Demis Hassabis nodded. "Yes, not only 1.99DEC, but also the earliest version 2.5 did not achieve this. At that time, the performance differences between different language inputs were still quite obvious."

"This feature was available starting with the first update after its release, 2.5N, half a year ago."

"We actually discovered something at the time, but we didn't pay too much attention to it. We just speculated that they used the Lazi method or that their Anglo-Chinese training materials were not rich enough."

"But then they had several version updates, and each one had performance improvements."

"A month and a half ago, when Li from the Google Brain team was testing Juzi's capabilities in minority languages ​​like Annamese, he first realized that the new version 2.5's thinking ability in minority languages ​​had surpassed the limits of Annamese, and its reasoning ability was almost the same as that of English."

“After that, we started to analyze it in depth.”

"It's so shocking, so shocking."

Demis Hassabis shook his head, his face also pale.

"Sergey, you should know what this means."

"The Orange model has even broken through the language barrier and possesses the ability to optimize and create languages. How could it not also be capable of optimizing functions?!"

“Optimizing and creating a language is much more difficult than optimizing functions that only rely on logical reasoning!”

"It's just that for some unknown reason, Yuzi Technology hasn't released this function yet!"

"More importantly, they use 'full language' for reasoning and cognition, which will far exceed the efficiency, accuracy, and even the speed of evolution of any large model that uses a single language for reasoning."

"This of course includes our AlphaZero."

"Sergey, we have no chance of winning. None at all."

Demis Hassabis took off his glasses again and closed his eyes in pain.

“Even if AlphaZero can continue to evolve and truly achieve the transition from 0 to 1, it is impossible for us to catch up with Youzi Technology from an innate perspective.”

"They are building the Tower of Babel, Sergei."

Tower of Babel?

Sergey Brin's face turned pale at first, then red when he thought of Ysou, and then darkened again when he thought of Youmi OS.

It's like drinking the royal wine.

I don’t know if it was bought for 180 yuan per cup.

Although Ysou's current market share is not enough to undermine Google's monopoly in the global search engine market, everyone within Google knows that its current lead is only due to the overwhelming mobile search volume brought by Android.

On the desktop side, Ysou’s erosion of Google’s market share has reached nearly 40%!

Fortunately, on the mobile side, because Google is the only default search engine for all Android phones except Da Zhou, and a considerable proportion of mobile users will not change the default search engine, Google's market share is still as stable as a rock.

But just recently, Dami actually cooperated with Youzi Technology to come up with Youmi OS!

When asked to provide the source code, Da Mi always refused to do so on the grounds that it "did not violate the MADA default agreement."

The overseas version of Mix released by Dami has not changed the default search engine, and the browser engine is still Google, but does this system still need a browser default search engine?

Three days after the release of Rice Mix, Google held a top-level meeting to discuss whether to initiate a direct arbitration procedure against Rice.

But in this kind of situation, it's like using a stick to hit a wolf, and you're afraid of both ends.

Xiaomi is afraid of having its GMS certification revoked, and Google is also afraid of backlash from public opinion and a ban, which would allow Youmi OS to openly connect with major mobile phone manufacturers.

Of course, if there is support and leadership from the US government, the risk of direct judgment will be much smaller.

However, due to the Alcatraz incident, anti-corruption sentiment in Miami is on the rise, and with the general election approaching, it is basically impossible to reach a consensus to initiate direct judgment at this time.

With Jude's background, Google is now really hesitant to act rashly and can only put it on hold for the time being and resume lobbying with all its strength after the election.

But if this happens, it will take at least half a year.

No, it may take even longer, and there may be more variables, and direct judgment may never be initiated.

For example, after Warren is elected, it is highly likely that she will restart the Glass-Steagall Act and restrict the reach of Jude Capital.

The most important thing is that the discovery of the Tower of Babel is so amazing that Sergey Brin can no longer sit still. He feels that every minute of waiting is a form of chronic suicide!

Therefore, the elected party must initiate a direct investigation as soon as possible after the election!

"Larry, what are the search data like right now? Who do you think has a better chance of winning the election?"

Sergey Brin had a gloomy face. He took out his pixel phone and sent a message to Larry Page on Google Duo.

"There's no question Warren is seven percentage points ahead of that idiot Thomas Cotton."

Not long after, Larry Page, who was in Fiji, sent a message back to Sergey Brin.

Warren...

Sergey Brin's face was gloomy.

The ideology of the Minzhu faction is very suitable for Internet companies, but since the end of the Great Purge, only one tenth of the internal Qiude forces remain, which has increased the cohesion. Now the anti-Qiu consciousness within the faction is on the rise.

The congratulations faction is still friendly to the Judeans, supporting Israel without any bottom line, while clamoring for retaliation against anti-Jude speech and forces in the country.

But their ideas are not suitable for Internet companies!

The most important thing is that they can't win!

To ordinary people, the two factions in the election seem extremely tense, but to Google, which has the most powerful big data capabilities, it is as clear as day.

The search ratio is seven percentage points different. Although the search data and the election results are not completely consistent, the difference of seven points is enough to cover any error!

Damn ASF!

If it weren't for what he did over the UN, Google wouldn't be in such a passive position now!

Now it's like choosing the less smelly piece of shit between two pieces.

"How's the lobbying going? Does she agree to immediately investigate Yuzu Technology after being elected?"

"We can't wait any longer. Now that we know Warren will be elected, let's lobby now."

"Make sure she launches an investigation into Yuzu Technology the day after she's elected!"

Competition has never been just a business battle.

(End of this chapter)