Chatbots based on language models have already become a part of today. But the most famous ChatGPT was launched only at the end of last year. And inside Google, a similar bot was offered much earlier, when the company’s management had a different view of AI.
According to the information by The Wall Street Journal, around 2018, Google research engineer Daniel de Freitas began work on creating a chatbot that could mimic human speech. Later, Noam Shazeer from the artificial intelligence research unit joined the project.
Together, they were able to build Meena, a chatbot that could chat about philosophy, TV shows, and make jokes about horses and cows. At the same time, as the developers’ colleagues said, they were sure that Meena would be able to fundamentally change the search on the Internet.
The bot later changed its name to LaMBDA and became the language model for Bard.
But before that, numerous attempts by engineers to provide access to external researchers, to integrate the chat function in Google Assistant or to launch a public demo guide were rebuffed. So the project was closed due to the fact that the chatbot, according to the managers, did not meet the standards of safety and fairness in the field of artificial intelligence.
And this is not the first time that similar AI projects have been discontinued at Google. In 2012, the company hired Ray Kurzweil, who was engaged in the development of language models. About a year later, the company acquired DeepMind, which also worked on AI. At the time, such initiatives were discontinued due to criticism from scientists and technical experts who were against the use of such technologies due to ethical problems related to mass surveillance, and Google promised to limit the use of AI.
In 2018, Google also canceled a project related to the use of AI in military weapons. This time, the negative reaction of the company’s employees played a role.
Of course, today the use of AI for Google played in a completely different light. But in any case, Bard will be out later than ChatGPT in Microsoft’s search engine, which is currently adding prices to the Redmond company’s stock. But Bard’s failed presentation already had its consequences for Google.
In late 2021, de Freitas and Shazeer left Google, despite CEO Sundar Pichai personally asking them to stay and continue working on the chatbot. After leaving Google, the engineers founded their own company, Character.Ai, where they are also working on a chatbot that can transform into characters like Elon Musk or Mario from the Nintendo game.