Tech’s longtime mantra — “move fast and break things” — has perhaps never been more applicable than it is now. The sprint for artificial intelligence supremacy is underway, with companies admitting they will get things wrong but still rushing to claim power in a new computing era.
At Google, that push for speed is reportedly running up against employee input, and employees are losing.
The Mountain View-based tech titan scrambled to turn out its ChatGPT competitor despite major employee concerns and prioritized speed over ethical commitments, according to a new Bloomberg report that relies on internal documents and interviews with 18 current and former workers. Reportedly, senior leaders overruled an AI governance team’s risk evaluation saying Bard wasn’t ready. Google opened Bard to the public in March as an “experiment.”
In February, one employee reportedly described Bard as “worse than useless,” adding that it should not be released to the public, in an internal post that was met with widespread agreement, according to Bloomberg. Others who tested the technology reportedly called out the chatbot for delivering factual inaccuracies and contradictory answers in response to simple questions.
Indeed, Bard’s launch was marred by mishap. A promotional video for the chatbot, which was built on Google’s history of artificial intelligence research, showed it giving a misleading answer to a factual question about the James Webb telescope. Google’s stock sank immediately after the incident — to the tune of more than $100 billion in market value.
Employees told Bloomberg that Google leaders decided to call new AI products “experiments” to help the public forgive any shortcomings. Bard’s interface currently says, “I’m Bard, your creative and helpful collaborator. I have limitations and won’t always get it right, but your feedback will help me improve.” Like other chatbots, in finer print at the bottom of the screen, it says that, “Bard may display inaccurate or offensive information that doesn’t represent Google’s views.”
With Bard launched and public for testing, Google is racing to upgrade its dominant search engine with AI features and build a new AI-powered search option, according to the New York Times. Google designers, engineers and executives are reportedly working in “sprint rooms” to test new versions — the new search engine would be much more personalized than the current search service.
“Responsible AI remains a top priority at the company, and we are continuing to invest in the teams that work on applying our AI Principles to our technology,” a Google spokesperson told SFGATE in an emailed statement.
Amid the technological push, Alphabet and Google CEO Sundar Pichai went on “60 Minutes” last Sunday for a concerned and somber interview that often focused on the dangers of charging forward with artificial intelligence technology.
“The development of this needs to include not just engineers but social scientists, ethicists, philosophers and so on,” Pichai said. But ethicists’ voices are getting boxed out of AI development at Pichai’s own company, according to Bloomberg’s reporting — a pattern that goes back to the high-profile exit of prominent AI ethicists Timnit Gebru and Margaret Mitchell in 2021.
Members of Google’s AI governance team reportedly submitted a risk evaluation saying Bard could cause harm and thus was not ready for launch. Google’s AI governance lead, Jen Gennai, said in a statement to SFGATE that reviewers don’t have final recommendations on product — she escalated the issue to senior leaders at the company, who approved Bard’s “experimental launch with continuing pre-training, enhanced guardrails, and appropriate disclaimers.”
Some employees told Bloomberg that they believe Google is conducting enough safety checks for its new generative AI products, but with quick releases prioritized, ethics employees said it can feel futile to speak out.
Hear of anything happening at Google or another tech company? Contact tech reporter Stephen Council securely at [email protected] or on Signal at 628-204-5452.
