Before ChatGPT’s artificial intelligence tool was released to the world, novelist Robin Sloan tested a similar AI writing assistant built by researchers at Google.
It didn’t take long for Sloan, author of the bestseller “Mr. Penumbra’s 24-Hour Bookstore,” to realize that the technology was useless to him.
“A lot of the state-of-the-art AI today is impressive enough to raise your expectations and make you think, ‘Wow, I’m dealing with something that actually works,'” Sloan said. “But in a thousand little ways, a million little ways, it lets you down and betrays the fact that you really don’t know what’s going on.”
Other companies may have released their experiments into the wild, as OpenAI did with its ChatGPT tool late last year. But Google is more cautious about who will play with its AI advances despite growing pressure for the internet giant to compete more aggressively with rival Microsoft, which is pouring billions of dollars into OpenAI and integrating the technology into Microsoft products.
The pressure is starting to ease, as Google has asked one of its AI teams to “prioritize working on a response to ChatGPT,” according to an internal memo reported this week by CNBC. Google refused to confirm that there is a public chatbot in the works but spokeswoman Lily Lin said it continues “to test our AI technology internally to make sure it is helpful and safe, and we hope to share more experiences externally soon.”
Some of the technological breakthroughs driving the field of generative AI – which can create paragraphs of readable text and new images as well as music and video – have been pioneered in Google’s research hands.
“So we have an important stake in this area, but we also have an important stake not only leading in being able to generate things, but also in dealing with the quality of information,” said Zoubin Ghahramani, vice president of research at Google, in a November interview with The Associated Press.
Ghahramani said that the company also wants to be measured in terms of what it releases, and how: “Do we want to be accessible in a way that people can mass produce things without control? The answer is no, not at this stage. I don’t think we will be responsible for we are the ones who drive.
And they don’t. Four weeks after the AP interview, OpenAI released ChatGPT for free to anyone with an internet connection. Millions of people around the world are now giving it a try, sparking lively discussions in schools and corporate offices about the future of education and work.
OpenAI declined to comment on the comparison with Google. But when they announced their full partnership in January, Microsoft and OpenAI said they were committed to building “reliable and secure AI systems and products.”
As a literary assistant, neither ChatGPT nor Google’s version of creative writing comes close to what humans can do, Sloan said.
The fictional Google is central to the plot of Sloan’s popular 2012 novel about a mysterious San Francisco bookstore. This is one of the reasons why the company invited him along with several other authors to try out the experimental Wordcraft Writers’ Workshop, which is powered by a powerful AI system known as LaMDA.
Like other language learning models, including the GPT line built by OpenAI, Google’s LaMDA can generate convincing text and talk to humans based on what it processes from a wide range of online texts and digital books. Meta’s parent Facebook and Amazon have also built their own large models, which can add voice assistants like Alexa, predict the next sentence of an email or translate languages in real time.
When it first announced the LaMDA model in 2021, Google emphasized its flexibility but also increased the risk of dangerous abuse and the possibility of copying and amplifying biased, hateful or misleading information.
Some Wordcraft writers find it useful as a research tool – like a faster and more definitive version of Google search – for asking for a list of “rabbits and magical qualities” or “verbs for what fireflies do” or for “Tell me about Venice in 1700 ,” according to Google’s paper on the project. But less effective as a writer or rewriter, turning boring sentences into clichés and showing gender bias.
“I trust them — they’re thoughtful and careful,” Sloan said of Google. “This is not a reckless technology model that rushes to launch this into any world.”
The development of this Google model has no internal problems. First, it excludes some important researchers who have examined the risks of the technology. And last year, he released an engineer who publicly posted a conversation with LaMDA in which the model falsely claimed to have human-like consciousness, with “various feelings and emotions.”
While ChatGPT and its competitors may never produce recognized literary works, the hope is that they will begin to transform other professional tasks – from helping to debug computer code to creating marketing pitches and speeding up the production of slide presentations.
This is the key to why Microsoft, as a seller of workplace software, is eager to improve its products with the latest OpenAI tools. The benefits are less clear for Google, which relies heavily on advertising dollars when people search for information online.
“If you ask a question and get the wrong answer, that’s not good for search engines,” said Dexter Thillien, a technology analyst for the London-based Economist Intelligence Unit.
Microsoft also owns a search engine – Bing – but ChatGPT’s answers are inaccurate and out of date, and the cost of running these queries is so high that the technology poses a serious risk to Google’s dominant search business, Thillien said.
Google has said that its previous big language model, named BERT, has played a role in answering online searches. The model can help generate a growing fact box next to the list of Google ranked web links.
Asked in November about the hype surrounding AI applications such as the OpenAI DALL-E image generator, Ghahramani admitted, in a cheerful tone, “sometimes it’s a bit annoying because we know we’ve developed a lot of the technology.”
“We’re not in this to get ‘likes’ and clicks, are we?” he said, he is listening that Google has become a leader in publishing AI research that others can build upon.
Learn how to navigate and strengthen trust in your business with The Trust Factor, a weekly newsletter that examines what leaders need to succeed. Log in here.