Imagine that your company has just hired a rising executive star so seductive that a competitor has just employed a lookalike. They are surrounded by an exhilarating buzz. Everyone, from the CEO to the shareholders, seemed to agree that this individual represents the company's future.
Then you hear the executive has what is nicely dubbed a "hallucination problem(Opens in a new tab)." Every time they speak, there is a 15 to 20% possibility that they will just make stuff up (Opens in a new tab). A professor at Princeton called the individual a bullshit generator(Opens in a new tab) (Opens in a new tab). They genuinely are unable to discern fact from fiction (Opens in a new tab). In five minutes, they will present a new product on stage. Do you still push them into the spotlight?
This week, Microsoft and Google responded affirmatively. Fired up by the success of OpenAI's ChatGPT, the Artificial Intelligence chatbot with 100 million monthly active users two months after its launch, Microsoft hosted a last-minute-surprise event to announce OpenAI would bring ChatGPT-style search to the Bing search engine and Edge browser. Google introduced its own AI search tool, Bard, and revealed it at a presentation in Paris the following day, but ran into its own hallucinatory problem.
"A new race starts today," Microsoft CEO Satya Nadella told reporters who'd been summoned to the Redmond, Wash. headquarters Tuesday. Yes, isn't it lovely to believe so? (Opens a new tab) Microsoft, the chronically uncool kid on the tech block, would love you to think that Bing — sorry, "the New Bing" — is in a contest with Google search on anything.
Google's initial announcement of Bard(Opens in a new window) reeked of condescension: "We re-oriented the organization around AI six years ago," wrote Google CEO Sundar Pichai.
The 'hallucination problem' and Google.
Which is a telling point. Google, the global search king, has had years to add AI, although its ChatGPT competitor, Bard, is barely in beta with a few of testers. Despite Pichai's hipster affectation, the Bard launch was unintentionally chaotic. Google also appears to have been taken off guard by the ChatGPT buzz.
How else to explain the awful Bard blunder on full show at launch – not during the event itself, when some demo flubs are expected, but in a pre-made GIF? A user is depicted asking Bard for information about the James Webb Space Telescope that he may share with his 9-year-old son.
One of the "facts", that the JWST took the first ever photograph of an exoplanet, is incorrect. Bard was delusional (Opens in a new tab). (UPDATE: While a Financial Times writer argues Bard's statements were technically true, this requires a reading of the language that no human would ever utilize – another issue with AI search.)
No surprise parent firm Alphabet's share price dropped as much as 8% on the day Bard was released. Google highlighted the primary issue with AI search and claimed that the firm cannot use its massive data repository to fact-check itself.
Google ought certainly know better, given that it had a "hallucination problem" in 2017 with its highlighted snippets at the top of search results. The snippets algorithm seemed to particularly like telling lies about U.S. presidents. Again, what could go wrong?
NEVER rely solely on a single source: The new guidelines for online learning
In other words, start your AI search tool too early and you're at risk of playing yourself. Microsoft got lucky, in the sense that no blatant mistakes were on display at its launch event. But if ChatGPT-based search weren't littered with blunders, why is it at such a shaky beta stage? If you are interested in performing unpaid AI quality assurance for Bing, there is a sign-up sheet (Opens in a new tab). https://ejtandemonium.com/
"There's still more to accomplish there," Sarah Bird, Microsoft's Head of Responsible AI (a revealing title!) said in answer to a question from Wired regarding ChatGPT's hallucination problem(Opens in a new tab) (Opens in a new tab). The 15 percent hallucination figure originated from a firm that is competing to develop a ChatGPT fact-checker (Opens in a new tab). (UPDATE: a New York Times columnist's frantic piece on New Bing discovered that it couldn't even perform basic math correctly or provide a list of kid-friendly activities in the surrounding area.)
Bird noted that past versions of the software may assist users in planning a school shooting, but this feature has been removed. Good to know! What could possible go wrong next? Surely, this hallucinogenic experimental search tool has no further unforeseen consequences that could humiliate a huge and legally vulnerable digital corporation.
Clippy. Zune. New Bing.
Microsoft knows from shame, of course: It's the firm that gave us one of the worst misfires in software history, Clippy. The paperclip assistant was famous for providing unwanted counsel. ChatGPT isn't Clippy, in the sense that we're coming to it with questions.
ChatGPT-enabled Bing may be compared to Clippy on LSD due to the fact that it frequently hallucinates its responses or, more often than you'd expect, provides users with a version on "I can't answer that." If enough casual users of "New Bing" encounter distorted results, then it will be its lasting legacy. http://sentrateknikaprima.com/
It makes little difference if a product improves over time; the initial response of the public is what turns it into a joke. Microsoft should know that, too; it gave us the Zune. It is the same to release a ChatGPT product before it is genuinely ready for prime time.
"The New Bing" is already begging to be a joke, to be honest. Or are you really ready to leave Google search and your Chrome browser for Bing and Edge, should the latter win the AI search competition, whatever "winning" truly means here? Didn't think so. As a factor, technological inertia is greatly undervalued.
ChatGPT is impressive in some situations — real estate brokers in particular love it for creating listings — and terrifying in others. Once one delves behind the headline, though, each account of its disruptions becomes diminished. It will result in a flood of student plagiarism! Except that it can also inform you when a paper has been written by ChatGPT, negating its own threat. It passed a law school exam! Except that it barely passed with a C-plus (Opens in a new tab).
Building the digital version of a human brain, termed as "generic AI" in the AI community, is extremely challenging. Another long-held aim for artificial intelligence is the development of insect intelligence, which has only just begun to be achieved. Will you actually rely on ChatGPT to provide your search results, as opposed to, you know, clicking on links yourself?
The answer could possibly depend on how much you yourself, my reader, are having a hallucination problem.