With battles for search engine, social media and smart assistant dominance largely fought and won, Big Tech is now turning its focus to chatbots. Given the scope of their potential, it is a fight that could overhaul all previous victories.
Possessing the world’s most dominant artificial intelligence bot could transform entire industries and disrupt society in ways so profound that we can’t even fathom them yet – at least according to those developing the technology.
The arrival of OpenAI’s ChatGPT late last year showed that it won’t necessarily be one of the established tech firms who make the big breakthrough – at least not publicly – but it will almost certainly be exploited by one.
Microsoft announced this week that it is already making use of its $10 billion stake in OpenAI to introduce new ChatGPT tools to its software services. Google’s AI division DeepMind also revealed this week that it is considering releasing its Sparrow chatbot this year.
Google also its LaMDA chatbot, though it has been reluctant to release it after one of its engineers claimed it was sentient. A certain amount of detachment, like with DeepMind or OpenAI, means tech giants can avoid risking their entire reputations.
It took Microsoft just 16 hours to learn the dangers of releasing an AI chatbot to the public, when its Tay bot on Twitter began to demean minorities and glorify Nazis before being swiftly shut down in 2016. A year later, Facebook made headlines when one of its AI bots appeared to invent a new language that humans couldn’t understand. Even IBM’s Watson, best known for beating human contestants on the gameshow Jeopardy, has proved tricky, with engineers fitting it with a swear filter after it memorised the Urban Dictionary.
Human-like chatbots are inevitable, but so too are the problems that will come along with them. The rush to monetise them in the current hype wave could mean issues are either overlooked or underplayed.
OpenAI’s technology may have built-in safety features, but that doesn’t mean it’s never wrong, or never racist. Expecting a bot to be able to figure out truth from lies, or understand the difference between an unintentional error and calculated misinformation, is a massive task, especially given how hard some humans find it
Lawsuits are already underway to limit AI tools, including ongoing EU regulation, though history suggests that trying to stop the advancement of technology is like trying to catch rain with a sieve.
DeepMind CEO Demis Hassabis describes artificial intelligence as an “epoch-defining technology, like the internet or fire or electricity”. Microsoft CEO Satya Nadella predicts generative AI will create “a world where everyone, no matter their profession” could use the tech “for everything they do”. Meanwhile, OpenAI boss Sam Altman says improvements to its chatbot will soon make ChatGPT “look like a boring toy”.
Like with social media or search engines, it may take a while to realise the potential damages or impacts of human-like AI. But while the outcome of the battle over chatbots may be uncertain, its conclusion may come fast.
