Vint Cerf, the “father of the internet” and top evangelist at Google, has offered a warning for company leaders wanting to close chat AI deals quickly and it sounds like “don’t.”
Vint Cerf cautioned against rushing to invest in conversational AI merely because “it’s a hot topic.” The warning comes as ChatGPT is experiencing a surge in popularity and Google is trying to take on the Microsoft-backed tech startup OpenAI with its Bard.
“There’s an ethical issue here that I hope some of you will consider,” he commented on Google’s Bard conversational AI.
“Everybody’s talking about ChatGPT or Google’s version of that, and we know it doesn’t always work the way we would like it to,” he said.
His caution comes amid tech firms’ struggle with maintaining competition in foolproof conversational AI.
John Hennessy, the chairman of Alphabet, stated that the systems are still far from being widely used and that several problems with “toxicity” and inaccuracy need to be fixed before the product is even tested on the general population.
Since 2005, Vint Cerf has worked for Google as a vice president and “chief internet evangelist”. As he collaborated on some of the architecture used to lay the internet’s groundwork, he is regarded as one of its “fathers.”
Although the technology is “really cool, even though it doesn’t work quite right all the time,” Vint Cerf advised against giving in to the desire to invest.
Vint Cerf remarked, “Man, I can sell this to investors because it’s a hot topic, and everyone will throw money at me.”
“Think carefully. You were correct in saying that we can’t always foresee what would happen with these technologies, but to be completely honest, the majority of the issue is people, which is why we as a species haven’t changed much over the past 400 years, much less the past 4,000,” he added further.
Vint Cerf stated, “They will seek to do that which is to their benefit and not yours. So, we must keep it in mind and exercise caution when utilising these technologies.”
According to him, there needs to be more clarity between what technology claims it can do and what it achieves.
“That’s the issue. You can’t tell the difference between an accurate and elegantly delivered response,” he said.
Vint Cerf gave an example when he asked a chatbot to provide him with a biography. He claimed that even though the bot’s response was inaccurate, it was presented as fact.
According to him, engineers like him should be in charge of attempting to discover a mechanism to control some of these technologies so that they are less likely to be harmful.
“A poor piece of fiction is one thing, depending on the application, of course. Offering someone advice can have medical repercussions. So it’s crucial to determine how to reduce the worst-case scenario,” he said.
Meanwhile, talking about the AI race, Google, which had a rocky start with its chatbot Bard in February 2023, is now implementing new methods to improve the chatbot’s efficiency.
As per Business Insider, the tech giant has already started companywide testing of Bard, with CEO Sundar Pichai asking his employees to spend two to four hours daily helping to test the chatbot before it gets integrated into Google search.
Google has started ‘dogfooding’ (testing by staff before it’s made publicly available) Bard since February 14, 2023. The chatbot reportedly has thousands of internal and external testers using it and submitting feedback.
After Bard’s launch, Google suffered a major reputational disaster as the chatbot offered incorrect responses to a question on the James Webb Space Telescope. This led to the search giant’s stock plummeting over 9%.