Former OpenAI research head says Google caught up after ChatGPT momentum faded
A former OpenAI vice president says the company squandered its early dominance after launching ChatGPT, allowing Google to rapidly close the gap and intensify competition in frontier AI.
Former OpenAI vice president of research Jerry Tworek believes that Google’s AI comeback was the result of OpenAI fumbling its lead. Tworek, who stepped down from his post earlier this month, shared his views on the company, AI research, and more.
Last year, Google narrowed the gap with OpenAI in the race to build frontier AI models. The Sam Altman–led startup that revolutionised consumer-facing AI with ChatGPT had dominated the segment since its launch. Lately, however, the search giant has matched pace and, according to some observers, outperformed OpenAI on model capabilities.
Tworek joined OpenAI back in 2019 when the company had around 30 staff. He is known to be the architect responsible for some major breakthroughs, including the advanced o-series reasoning models. Even though the former OpenAI executive did not elaborate on what has been the company’s biggest misstep, he asserted that the company wasted its massive edge over others that it gained after launching ChatGPT.
According to him, if you own a company that is ahead and has all the advantages that OpenAI has, you should always stay ahead. Tworek shared that on the other hand, Google went on to do so many things right. “Very clearly, Google started treating it seriously at that moment, training large language models, and, like, through OpenAI fumbling its lead, they are very, very close now in capability and in terms of models trained,” he said in the interview by YouTube channel Core Memory Podcast.
Tworek’s comments come at a time when OpenAI is facing intense competition as well as scrutiny for its stand on user privacy. The executive also believes that aggressive competition in AI has essentially changed how AI research is conducted. According to him, the pressure to push user growth and pay for computing resources while at the same time competing for the best model has reduced the risk appetite of companies for bold research that may not offer instant results. He feels that artificial general intelligence will require radical approaches, and it may arrive by 2029.
Earlier this week, at Davos, while speaking at a session hosted by the All-in podcast, Microsoft CEO Satya Nadella described the competition in AI as ‘pretty intense’. He said that he is glad about it and that we should look at competition as a percentage of the GDP five years from now.
Similarly, DeepMind CEO Demis Hassabis described the AI landscape as the ‘most intense environment’ veteran tech professionals have ever seen.
Former OpenAI vice president of research Jerry Tworek believes that Google’s AI comeback was the result of OpenAI fumbling its lead. Tworek, who stepped down from his post earlier this month, shared his views on the company, AI research, and more.
Last year, Google narrowed the gap with OpenAI in the race to build frontier AI models. The Sam Altman–led startup that revolutionised consumer-facing AI with ChatGPT had dominated the segment since its launch. Lately, however, the search giant has matched pace and, according to some observers, outperformed OpenAI on model capabilities.
Tworek joined OpenAI back in 2019 when the company had around 30 staff. He is known to be the architect responsible for some major breakthroughs, including the advanced o-series reasoning models. Even though the former OpenAI executive did not elaborate on what has been the company’s biggest misstep, he asserted that the company wasted its massive edge over others that it gained after launching ChatGPT.
According to him, if you own a company that is ahead and has all the advantages that OpenAI has, you should always stay ahead. Tworek shared that on the other hand, Google went on to do so many things right. “Very clearly, Google started treating it seriously at that moment, training large language models, and, like, through OpenAI fumbling its lead, they are very, very close now in capability and in terms of models trained,” he said in the interview by YouTube channel Core Memory Podcast.
Tworek’s comments come at a time when OpenAI is facing intense competition as well as scrutiny for its stand on user privacy. The executive also believes that aggressive competition in AI has essentially changed how AI research is conducted. According to him, the pressure to push user growth and pay for computing resources while at the same time competing for the best model has reduced the risk appetite of companies for bold research that may not offer instant results. He feels that artificial general intelligence will require radical approaches, and it may arrive by 2029.
Earlier this week, at Davos, while speaking at a session hosted by the All-in podcast, Microsoft CEO Satya Nadella described the competition in AI as ‘pretty intense’. He said that he is glad about it and that we should look at competition as a percentage of the GDP five years from now.
Similarly, DeepMind CEO Demis Hassabis described the AI landscape as the ‘most intense environment’ veteran tech professionals have ever seen.
