The head of Google talks about AI: Why is it “moving slowly”? Are there any concerns about artificial intelligence?

Source: Financial Associated Press

Author: Shi Zhengcheng

In early September, Google (Alphabet), an American technology company founded in 1998, celebrated its 25th anniversary. Group CEO Sundar Pichai published an enthusiastic open letter on the official blog, and Screaming that Google’s exploration of artificial intelligence technology will drive extraordinary technological progress in the next 25 years.

On the bright side, in the past 25 years, Google has had 15 products with more than 500 million users, and 6 products with a user base of up to 2 billion people. This one company alone** accounts for 40% of global digital advertising revenue**.

But from the perspective of September 2023, this company with a market value of US$1.7 trillion also faces many challenges: the search engine business on which the company relies for survival is about to face the impact of antitrust lawsuits. At the same time, Google, which has been deeply involved in the field of artificial intelligence for a long time, was upstaged by OpenAI. Microsoft even added large models to Bing to challenge Google's basic market.

In response to this series of questions, Alphabet and Google CEO Pichai gave detailed answers in an exclusive interview with "Wired" released at the beginning of this week.

**Question: Is Google moving too slowly in the field of AI? **

Although Google began exploring the field of artificial intelligence very early, even the Transformer paper that inspired OpenAI was written by Google researchers. But in this wave of AI productivity, Google has always appeared as a chaser.

Pichai emphasized that Google has been thinking about opportunities in the field of artificial intelligence. When he became CEO in 2015, deep neural networks were already obvious, so he also launched an "artificial intelligence first" strategy. Google has also considered launching large-scale consumer AI products, but the company is indeed more cautious in this regard.

"You can imagine, when a user uses Google to ask, 'How many doses of fever-reducing medicine should be used for a 3-month-old baby?', the responsibility for the correct answer means," Pichai said as an example.

The head of Google also praised OpenAI, pointing out that ChatGPT demonstrated the fit between the product and the market. This is also an exciting time for Google. However, he once again emphasized that "you need to be cautious" because the development of technology is a long-term process, and he is also very confident about Google's current position.

As for the statement that "Google has talent and technology, but was preempted by OpenAI", Pichai retorted that for the next 5 to 10 years, this order is not important. The fact is that when people see big languages Now that the model works, there's a lot more Google can do**. Pichai emphasized the need to distinguish “signals from noise.” The signal here is that artificial intelligence is a profound platform shift that is at a stage where it can be applied more deeply. **

**How to deal with the transformation of search engine advertising in the AI era? **

Similar to Bing, Google also launched SGE (Generative Search Engine) this year, but this also means that if users’ search habits change, the digital advertising business model that supports almost the majority of Google’s revenue will face challenges.

However, Pichai has a different view on this. He pointed out that in the AIGC search engine, Google will also provide users with website links from a series of sources, and the company also hopes that users can use these links, so the core experience part is not Nothing will change. Early experiments show that the company can deliver advertising space in a way that suits users. In fact, when the mobile Internet wave took off, there were similar doubts. He believes that Google can do better during the current transition period.

**How do you view antitrust lawsuits? **

Pichai said that this case occurred in an era of unprecedented innovation. Looking at the current wave of AI, there is a large number of software and options for people to obtain information. Google makes thousands of changes every year to improve the search experience. He looks forward to being able to prove that in court.

The head of Google emphasized that people's access to information is a very dynamic and vast field, and the company needs to work very hard to continue to innovate in order to stay at the forefront.

Emphasis on continuing to work closely with NVIDIA

Although Google has been developing and using self-developed AI chips (TPUs) for many years, Pichai avoided the topic of "Nvidia's market dominance" and emphasized that Google will continue to cooperate with it in depth.

Pichai emphasized that Google and Nvidia have been cooperating for more than 10 years, including in-depth cooperation on Android systems. Many Google Cloud users are also Nvidia users, so maintaining this collaboration is important. The semiconductor industry is a very dynamic and cooperative industry that requires long-term intensive research and development and large capital investment. Pichai is satisfied with the cooperation between Google and Nvidia and believes that such close cooperation will continue in the next 10 years.

**What long-term concerns does Google CEO have about artificial intelligence? **

In Pichai's eyes, artificial intelligence risks are divided into three types: short-term, medium-term and long-term, and resources must be invested in a balanced manner according to the stage.

He said that in the near future, the most prominent risk is the hallucination problem of large language models - robots will talk nonsense. This is appropriate in some creative fields (such as naming a dog), but not in an answer to "Dosage of medication for children." So right now, the most important thing is to test model security and ensure privacy protection and avoid introducing bias.

In the medium term, Pichai is most worried about whether artificial intelligence will replace or enhance the labor force. In some areas, AI will be a destructive force. In the long term, developing powerful intelligent agents will become a more important question - can "their" values be consistent with humans? How do we control "them"? **

Regarding the recently released movie "Oppenheimer", Pichai said that he is currently reading the original work, and his habit is to read the original work before watching the movie.

Regarding the theme of the film, Pichai also extended that everyone, including Google, who is engaged in powerful technology research to some extent - artificial intelligence, gene editing, etc., must be held accountable.