Google AI chatbot Bard gives wrong answer in its first demo
Wipes $100bn from Alphabet stocks
Google parent Alphabet saw a $100 billion decline in its market value on Wednesday after its AI-powered chatbot Bard gave incorrect answer in a promotional video, fuelling concerns that Google is falling behind competitor Microsoft.
Google unveiled its AI chatbot technology called "Bard" on Monday, claiming it would provide "fresh, high-quality responses" to users' queries by drawing on information from the web.
The company also shared a short GIF video of Bard in use on Twitter.
In the video, Bard is asked: "What new discoveries from the James Webb Space Telescope (JWST) can I tell my 9-year old about?"
Bard offers many responses, one of which claims that the JWST was used to capture the first images of exoplanets - the planets beyond the Earth's solar system.
However, the fact is that the first photograph of an exoplanet was captured by the Very Large Telescope (VLT) of the European Southern Observatory in 2004.
Several astronomers pointed out Bard's wrong answer on Twitter.
Grant Tremblay, an astrophysicist at the US Center for Astrophysics, wrote: "Not to be a ~well, actually~ jerk, and I'm sure Bard will be impressive, but for the record: JWST did not take 'the very first image of a planet outside our solar system'."
Bruce Macintosh, the director of University of California Observatories, stated: "Speaking as someone who imaged an exoplanet 14 years before JWST was launched, it feels like you should find a better example?"
Google said the error highlighted the need of testing new technologies.
"This highlights the importance of a rigorous testing process, something that we're kicking off this week with our Trusted Tester program," a Google spokesperson said.
"We'll combine external feedback with our own internal testing to make sure Bard's responses meet a high bar for quality, safety and groundedness in real-world information."
Many experts believe the hurried release of Bard and lack of detailed information about it are signs of Google's "code red" alert, triggered by the launch of ChatGPT.
ChatGPT is a rival chatbot from Microsoft-backed OpenAI that has taken the internet by storm since its debut in November last year.
Although the underlying technology of ChatGPT is not ground-breaking, OpenAI's choice to make the system freely accessible on the web has exposed millions of people to this innovative form of automated text generation over the last three months.
Google's Bard chatbot is powered by LaMDA, a language model built on Transformer - a neural network architecture. Interestingly, ChatGPT is based on the GPT-3 language model, which is likewise built on Transformer.
Google now wants to move quickly on AI chatbot technology to keep pace with OpenAI and other competitors. The company is reportedly bringing in founders Sergey Brin and Larry Page to expedite its efforts.
AI-driven search is appealing simply because it can provide results in plain English as opposed to a list of links, which could speed up and improve web surfing.
On the most recent earnings call, Pichai said that the world is now ready for generative AI.
"I feel comfortable with all the investments we have made in making sure we can develop AI responsibly and we'll be careful," he noted.
Despite that, Google is yet to provide any specifics regarding how and when the company plans to incorporate Bard into its core primary search function.
Commenting on the AI arms race unleashed by ChatGPT, Victor Botev, CTO at scientific research software specialist Iris AI, said more tailored language models can lead to greater factual accuracy for AI results.
"LLMs represent endless possibilities, but their turn in the spotlight exposes unsolved questions: factual accuracy, knowledge validation, and faithfulness to an underlying message.
"Big corporations like Microsoft have the resources and technological heft to tackle some of these issues for general search results, but this won't work for everybody. To solve these issues for niche domains and business applications requires substantial investment. Otherwise, these models are unusable.
"Every organisation operating in a niche domain needs accurate results that understand the specificity and idiosyncrasies of said domain. LLMs are not, and will not be, able to capture these nuances within the next couple of years.
"Instead of getting caught up in the generative AI craze that will dominate 2023, businesses and large tech corporations should consider the AI technologies that will drive real value, rather than driving headlines. Bigger does not always equal better."