From OpenAI to Anthropic, is it the fate of all AI labs to be invested or acquired?

According to reports, Google will invest $300 million in Anthropic. While details about the deal between Google and Anthropic are sparse, it bears many similarities to Microsoft's investment in OpenAI, which recently received a multi-year, multi-billion-dollar investment.

As the race for the next dominant position in computing escalates, it’s hard for artificial intelligence research labs like Anthropic to avoid getting caught up in the market competition between tech giants. The crux of the matter is: How do AI research labs strike the delicate balance between pursuing scientific advancement and creating commercialized technology?

Anthropic "born out of OpenAI"

Anthropic was founded in 2021 by Dario Amodei, former vice president of research at OpenAI. The company doesn't make its work public beyond publishing research papers, but it has been very successful in raising funding for its research. The company raised $124 million in Series A funding when it was founded, and received an additional $580 million in Series B funding last year. Now it has received another US$300 million investment from Google.

Anthropic describes the company on its website as an AI security and research company dedicated to building reliable, explainable and controllable AI systems. The company is trying to make progress on today’s problem of “unpredictability, unreliability, and opacity in general-purpose AI systems.”

Anthropic has published more than a dozen research papers on its website on topics including Large Language Models (LLM), Reinforcement Learning from Human Feedback (RLHF), Large Generative Models, Transformer Models, and Interpretability sexual content. Its most recent paper was published in December 2022.

Anthropic has no public plans to launch commercial AI products and is currently focusing on technical research. But the company said, "On the road ahead, there will be many opportunities for our research to create value for commercial and public interests."

Google and Anthropic's "natural match"

Anthropic has selected Google as its preferred cloud computing provider to train, scale and deploy its AI systems. Dario Amodei said, “Google Cloud’s investment will support Anthropic’s next phase of development, deploying our AI systems to a larger user base.”

The company is developing an AI assistant called Claude, but it has not yet been released to the public (but some media have compared it to OpenAI’s ChatGPT). Claude uses a self-developed technology called "Constitutional AI" that can work with other AIs to supervise the behavior of AI systems. This technology aims to control AI behavior more precisely and reduce human intervention.

Google already has some of the most talented AI scientists in its Google Brain and DeepMind divisions, so why would it be interested in working with Anthropic? There may be three key reasons for this, all of which run parallel to the collaboration between Microsoft and OpenAI.

First of all, Microsoft's investment in OpenAI has demonstrated the benefits of investing in Anthropic rather than acquiring it. Although generative AI is exciting, it is an emerging field. Developers are still exploring its ethical, legal and technical challenges. Some lawyers are still trying to figure out how models like DALL-E and Codex affect copyright infringement laws. The developers of GPT-3 and ChatGPT are continually creating new security measures to prevent large language models (LLMs) from generating hate speech and other harmful content, while educational institutions are concerned about students using generative models to cheat on exams.

For big tech companies like Google and Microsoft, any of these concerns could turn into a PR nightmare and adversely affect their brand and reputation, so the likes of OpenAI and Anthropic Startups need more room for trial and error and learning.

Secondly, Anthropic has become a long-term customer of Google Cloud Platform. Google is investing in GCP credits rather than cash, which will likely make Anthropic a long-term Google customer.

Third, partnering with Anthropic will allow Google to learn to improve its AI hardware to run models at scale. Google already has an experienced AI hardware engineering team, and its TPU processor is one of the best inference engines of its kind. But the Anthropic team’s participation in OpenAI’s “AI & Compute” research work will definitely bring new ideas.

The future of Anthropic AI research

One question that industry people are concerned about is, "Where will the partnership between Google and Anthropic go?" Industry experts believe that Google will become a bigger The winner, because Google wins at least one high-value customer and prevents this promising AI research laboratory from being swallowed up by Microsoft's growing applied AI market. And in the best-case scenario, Google will consider acquiring Anthropic in the future to compete with other technology giants.

So will Anthropic suffer? The company received the funding it needed to continue research, but would find itself locked into Google Cloud Platform because they received GCP credits. Once their scalable AI infrastructure matures, switching it to a different provider will be difficult and costly.

Considering their commitment to expand their business, even the $300 million in cash may be exhausted after a year. After that, they will have to raise more money to cover their operating costs on Google Cloud. With any luck, Google will continue to provide more GCP credits. Otherwise, Anthropic will have to find a way to monetize their technology, and who better than Google?

This is too similar to the development trajectory of OpenAI since receiving Azure credit investment from Microsoft. place. While OpenAI continues to publish stunning research results, it is also influenced by the commercial interests of its major financial backers. It is actually more focused on research that is consistent with Microsoft's product line and can be monetized faster (such as language models, text conversion image generators, source code generators), rather than working on research that has purely scientific purposes and cannot be achieved in the short term. Realization fields (manipulators, game robots, evolutionary algorithms, etc.).

Before founding Anthropic, Amodei and other researchers reportedly left OpenAI after receiving a $1 billion investment from Microsoft due to disagreements over the company's development direction. They were worried that with Microsoft's investment, OpenAI would turn to a commercial path and eventually deviate from its goal of focusing on AI security.

Now, Anthropic has taken the first step down a similar path. Will this AI research lab stick to its scientific goals while receiving funding from big tech companies? Or will it eventually become another de facto branch of Google, creating technology that can be integrated into Google products? Where to go from here?

The above is the detailed content of From OpenAI to Anthropic, is it the fate of all AI labs to be invested or acquired?. For more information, please follow other related articles on the PHP Chinese website!