Noam Shazeer, a key figure in the development of Google’s large language AI technology, believes the company long held back from releasing a chatbot out of worry it could backfire. Shazeer, a former Google Brain engineer, is now the CEO of Character.ai and recently discussed his new startup on the “No Priors” podcast. Character.ai, a bourgeoning generative AI-based company, has raised almost $200 million to date to provide “characters” which are AI-backed and resemble personalities from celebrities, therapists to life coaches.
The large language research conducted by Google had the potential to develop a chatbot years ago. Shazeer and Daniel De Freitas, President of Character.ai, who also came from Google Brain, co-founded the Character.ai. De Freitas, being an ambitious individual with a vision to set chatbots as reality, joined Google in 2016. After working as a “20% project” and recruiting a team who also dedicated their working hours to help him, De Freitas developed Meena, a chatbot that was then renamed LaMDA.
Google, however, had its doubts over the venture of releasing a bot. De Freitas recognized the potential of language technology and saw the progress. But the search giant was more cautious of the impact due to fear of any wrong saying and the potential reputational damage that could follow.
The idea for a chatbot has been thrust back into the public sphere after OpenAI introduced ChatGPT. Microsoft capitalized in the idea by investing billions into OpenAI and integrated the technology into its Bing search engine; Google soon followed by introducing Bard.
Unexpectedly, the controversy was sparked when Blake Lemoine, an engineer, proclaimed LaMDA was conscious and needed human rights. His statement was soon followed by pushback from AI researchers like Timnit Gebru and further speculation that chatbots could potentially drive visitors away from copyright-protected websites also emerged. Google had to work extensively on training Bard to produce verified answers and protect users from any potential harm.
Today, companies like Character.ai are making immense progress in the field of generative AI, taking steps to guarantee chats are not sexually suggestive. Character.ai’s technology works off a large amount of text-scraped information from the web in order to obtain its knowledge. Google, Microsoft, and OpenAI are all battling to possess the title of pioneers in the world of AI-based chatbot technology, seeking to drive traffic to their services and gain market dominance.
