Shulou(Shulou.com)11/24 Report--

Faced with the chaos of GPT's crazy marketing, Google researchers issued a long post against its commercialization model: hype can not save GPT! But the investor said: it doesn't matter, just have fun.

In the field of AI, the king of traffic last year was generative AI.

From DALL-E 2 to ChatGPT, from Stable Diffusion to Midjourney,AIGC industry shows explosive growth.

A steady stream of generative tools based on artificial intelligence have dazzled investment companies and the general public.

Sequoia Capital, a top investment company, wrote in a blog post: "generative artificial intelligence is not only faster and cheaper, but also in some cases better created than humans. It is easy to imagine that decades from now, generative artificial intelligence will be deeply rooted in the way we work, create and play. "

Investment companies' enthusiasm for AIGC can be measured in terms of money, with 78 deals in the sector estimated at $1.37 billion in 2022 alone.

However, while the creators and investors of the AI model believe they are the power to change the planet, not everyone in this field believes that these generation machines are the best choice.

Google researcher: hype can't save GPT. "the current atmosphere of artificial intelligence has many similarities to the web3 craze in 2021, which makes me uncomfortable." Fran ç ois Chollet, a Google deep learning expert and founder of the deep learning system Keras, issued a strong warning on Twitter about the current investment boom. "people believe stories that don't tell the truth. "

Fran ç ois Chollet "everyone firmly believes that GPT will have a" civilization-changing "impact (and a 100x return on investment) in the next 2-3 years," he continued. "personally, I think there are bull markets and bear markets in the development of LLM. "

He believes that in a bull market, "generative AI has become a broad example of interacting with most technology products." Even in this case, general artificial intelligence (AGI) is still a pipe dream.

At the same time, in a bear market, large language models like GPT-3 will find "limited success stories in SEO (search engine optimization), marketing and copywriting" and turn out to be an "out-and-out bubble".

Chollet believes that the eventual development of LLM is likely to be somewhere in between and in favor of the latter. "artificial intelligence as a universal interface to our information" is certain to happen in the future, but this generation of technology has not been able to fully achieve this.

Commercialization is an important criterion for the success of the LLM model. If LLM can generate huge financial returns, then the commercialization of the technology will be considered a success.

OpenAI, for example, made a profit of about $5 billion to $10 million in 2021 and $3, 000 to $40 million by 2022. Even if it is as powerful as OpenAI, only the commercial value of image generation technology has been recognized by the market.

Chollet says it likes to search Twitter for the most popular ChatGPT tweets to learn more about the use cases of the technology. After browsing a large number of tweets, Chollet found that 80% of the posts were about how to drain, or even a gimmick to swindle the number of hits.

Whether it is the "10 Secrets of paying to unlock ChatGPT" or all kinds of ChatGPT training courses, the emergence of LLM has brought a subversive impact on the commercial model of traffic realization.

However, Chollet believes that the real potential of ChatGPT is much more than that. It is likely to shine in consumer goods, education and search.

No matter what happens in the future, people will soon know. Billions of dollars are pouring into the track to apply ChatGPT or similar technology to a large number of products. By the end of this year, people will have enough data to make judgments.

Of course, Chollet also said that despite the hype, deep learning can "build a lot of cool things." It was so five years ago, it is today, and it will still be so five years later. Even if the halo given to it by the author is removed, this technology is still very valuable.

Web3 cannot be compared with LLM. After all, web3 is a pure gimmick, and LLM is a real technology with practical applications. The hype mentioned by Chollet is a bubble formed among venture capitalists.

At the end of the article, Chollet explains how GPT technology is "marketed successfully". "the driving force behind investing in GPT is not experimental data or profit statements, but pure hype and baseless narratives. They form a self-consistent cycle: hype attracts investment, and an increase in investment leads to more speculation, which in turn leads to more capital inflows. "

Most importantly, it becomes common sense when a lie is repeated a hundred times. The "big cake", which is not supported by any data, has become a self-evident standard in word of mouth.

Marcus: don't be too sure that Chollet is not the only one in GPT who is cautious about GPT. Marcus, a professor at New York University, often throws cold water on the enthusiasm of the world.

In an interview, Marcus said that while ChatGPT seems to know everything, it is also error-prone. ChatGPT, as before, the relevant systems "are still unreliable, still do not understand the real world, still do not understand the psychological world, and are still full of errors. "

So while the AI world was delighted with the arrival of GPT-4, Marcus made seven less positive predictions.

1. GPT-4 will still make all kinds of stupid mistakes like its predecessors. Sometimes it does a good job of accomplishing a given task, sometimes it goes on strike, but you can't predict in advance what's going to happen.

2. GPT-4 's reasoning on physics, psychology and mathematics is still unreliable. It may be able to solve some of the previously unsuccessful projects, but it is still at a loss in the face of longer and more complex scenarios.

For example, when asked about medical questions, it either refuses to answer or occasionally utters nonsense that sounds reasonable but dangerous. Although it has engulfed a lot of content on the Internet, it is not credible and complete enough to provide reliable medical advice.

3. Fluency hallucinations (fluent hallucinations) will still be common and easily induced. In other words, large language models are still a tool that can be easily used to produce information that sounds reasonable but completely wrong.

4. GPT-4 's natural language output is still unable to provide services to downstream programs in a reliable way. Developers who use it to build a virtual assistant will find that they cannot reliably map the user's language to the user's intention.

5. GPT-4 itself will not be a general artificial intelligence that can solve any task. Without outside help, it can neither beat Meta's Cicero; in Diplomacy nor drive a car reliably, let alone drive Optimus Prime in Transformers, or be as versatile as Rosie in the Jason family.

6. The "docking" between what humans want and what machines do is still a key and unsolved problem. GPT-4 will still be unable to control its output, some suggestions are surprisingly bad, and examples of masking biases will be discovered within days or months.

7. When AGI (General artificial Intelligence) is implemented, large language models like GPT-4 may be part of the final solution, but only part of it. Simple "extension", that is, building a larger model until it absorbs the entire Internet, will prove to be useful to some extent. But trustworthy universal artificial intelligence, which is consistent with human values, must come from more structured systems. It will have more built-in knowledge and include clear reasoning and planning tools. And these are what the current GPT system lacks.

Marcus believes that within a decade, perhaps less, the focus of artificial intelligence will shift from the expansion of large language models to combining with a wider range of technologies.

Cool stuff is always fun, but that doesn't mean it can lead us to credible general artificial intelligence.

Investor: but it's fun, but even so, even if AI experts take great pains to persuade investors to "run away", GPT followers still have checks to express optimism about technology.

Just today, Microsoft is talking about investing $10 billion in OpenAI, which will increase the company's market capitalization to nearly $30 billion.

"this is the paradigm shift we've been waiting for," Niko Bonatsos of venture capital firm General Catalyst said in an interview. "" maybe it's bigger than we thought. "

In the eyes of investors, these algorithms are cool. The text-image generator is impressive and opens the door to the creative world for rookies who do not understand Photoshop. For them, GPT is at least fun to play with.

Although the industry CEO is open to the fact that these projects are still in a relatively early stage, and even if the future of the field is bright, for now, their destructive potential and blurred creative boundaries are hard to ignore.

In Chollet's view, to really achieve a "paradigm shift", products need to be not only cool and interesting, but even useful for niche products.

He warns that venture capitalists are taking a much bigger risk than they think, fueling the hype cycle of half-baked products and being frenzied with wool. Capital has not made a prudent forecast for this promising but still limited emerging market.

"everyone is starting to believe that they are the chosen ones, especially those who claim to be contrarian investors," he said.

Reference:

Https://futurism.com/deep-learning-expert-gpt-startups-rude-awakening

Https://twitter.com/fchollet/status/1612179805298135040

This article comes from the official account of Wechat: Xin Zhiyuan (ID:AI_era)

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.