Unprecedented popularity means great advertising opportunities.

At this time, Meng Fanqi was intensively testing the performance of the new face algorithm and the success rate of the generative fake face recognition in his company.

Originally, this only had to be done more or less, but now that the security of generative content is a big concern, there is no room for error in this regard.

While Meng Fanqi is refining the effects of his algorithms, facial recognition teams around the world are thinking hard about it.

Opportunities and challenges coexist, although the entire face recognition direction is facing a huge problem of difficult to distinguish between true and false, but as long as this problem is solved, the benefits are equally huge.

This kind of security hazard is a fundamental problem that is bound to be solved, and it can be said that whoever can solve this problem will master the core technology and be able to pick the biggest fruit one step ahead.

Facebook's DeepFace team had just iterated on several new data from Meng Fanqi, and before they could recover from the joy of making a few key improvements, they directly encountered a major crisis.

The brothers were numb.

A second ago, he felt that Meng Fanqi was the savior of his team, but now he felt that Meng Fanqi was the team's greatest enemy.

"How can this birdman make a technological breakthrough so quickly....." Yang Ming sighed deeply, a team of four people focusing on human faces on his side is familiar with experiments, and the division of labor and cooperation are even more intimate.

Why is it that every time I get close to being released, I am always interrupted by this birdman.

The last time was a revolutionary breakthrough in the backbone network, which led to several people having to iterate again to ensure the status of their technology.

It's good now, don't make a breakthrough, just change the whole track, and give you a revolutionary security problem.

This is a big bad luck, and it directly becomes a question of whether this technology can still be used.

Wiped out an entire track.

"How accurate is our algorithm in recognizing generative prosthetic faces?" Wolf asked.

"It depends on the situation, it mainly depends on how much proportion of your fake face you show it."

"How's that?" Wolf was stunned for a moment.

"Because it will think that all the faces are real anyway, so if you don't show it a fake face, the accuracy rate is 100%, and if you want to show it all fake faces, then one person will not recognize it."

"You're still poor, aren't you??" Wolf's face darkened, when was he still shaking his wit here.

"Alas. If the security issue cannot be solved, the company's plan to commercialize facial vision algorithms in the short term will be postponed into a long-term one. ”

Mrs. Wo is also numb, and the short-term commercial plan becomes long-term, which directly means that the value and status of their foursome are broken.

Researchers didn't expect this wave to reach the pinnacle of life, but they also wanted to improve their lives, but now it's far away.

Although companies like Facebook and Google are not stingy with technology investment, whether the technology you make can be monetized, the treatment and status are still much worse.

You are like Meng Fanqi, who was originally an ordinary worker in Google's brain, with a wave of the axe on three wheels, it is inconvenient for Google to give him so much share of cash flow, and it has been talked about a considerable part of it in the form of equity.

In an instant, he became a shareholder with a large share, and his status naturally rose.

This thing is outrageous enough, people haven't officially joined the company, and they have accidentally changed from a wage earner to a small boss.

Although converted to less than 0.3% of Google's equity that Meng Fanqi can get a year later, Google is a giant that has gone through multiple rounds of financing.

Even founders Page and Brin only hold about 5.5 percent of the shares.

Google's stock price has taken off all the way, from about 30 in '14 to nearly 150 at its peak in '22.

If billions of dollars in assets need to be preserved, it is also a good choice to hold them first.

When there is an investment opportunity, take out a part, such as Tesla and Bitcoin, which will be dozens of times the profit.

But these opportunities may not be able to withstand too much money, Musk will not sit back and watch himself crazy to buy the bottom, BTC itself buys too much, and capital may also choose other currencies to harvest the people.

In short, the aim is to first preserve value with Google and then diversify your investment. BTC, Tesla, invest in start-ups, start your own business, and blossom in more places.

That's the safest thing to do.

At the same time, at the Hong Kong Chinese University in Huaguo Port City, Chinese scholar Tang Xiaoou led a group of his students and researchers, with a solemn expression and a gloomy atmosphere.

Just a week ago, Tang Xiaoou's team's face algorithm recognition capability reached the human eye level (97.53%) for the first time, and according to the original plan, the team will spend two or three months to continue to advance this result.

Then it will open the era of face recognition technology.

It's a pity that the sky didn't fulfill people's wishes, and Meng Fanqi stepped in sideways, doubling the difficulty of landing.

"Is there any way we can tell the difference between generative content at the moment?" Tang Xiaoou is the mentor of Kaiming, the original author of the residual network, and Kaiming's amazing dark channel defogging algorithm in 09 was made under the guidance of Professor Tang.

It won the best paper of the CVPR of the Vision Conference, which is the first Asian paper to win this honor in the 25-year history of the conference.

Before deep learning became so popular, Professor Tang's lab had been publishing papers at international summits.

In the 11-13 years of the top vision conferences, Professor Tang's lab has produced as many papers in the field of deep learning as all other institutions in the world combined.

It is enough to show its vision, technical strength and theoretical background.

But Professor Tang is still at a loss in the face of this sudden security problem.

"Mr. Tang, although FaceGAN has published the paper and the display website, we are completely unclear about the details of its specific training code and practice. In this case, it is difficult for us to conclude any valuable laws. ”

"Yes, Mr. Tang, generative technology is a relatively new technology, and it is the first time it has been used in face images. We have no experience in this area, and it may take quite some time to solve this problem. ”

"If we were to collect a lot of generated content through their [these people don't exist] website, would there be any chance of drawing any patterns?" Tang Xiaoou frowned, the other party's technical path was novel, and he was completely unfamiliar with it.

It's been too long to figure it out at the source.

His idea is to make a targeted detector specifically through these fake contents.

It is carried out independently from the original core recognition part of oneself.

It has to be said that this is also a way of thinking, and the models that can generate these fake data are also trained from this kind of confrontation, and it is difficult to achieve any results in the short term.

"We'll release the technology first, and then we'll solve the problem." Almost at the same time, both Facebook and Tang Labs made the decision.

In the original timeline, the technical release on both sides would have been delayed by another month or two.

But the sudden problem made both sides realize that there are many dreams at night, and if they don't release it, the situation will be even more embarrassing.