Microsoft gambles with the truth: Is ChatGPT just a lying kid?

Is Google losing its monopoly in the search engine market? Microsoft wants to find out with its new Super Bing.

Microsoft gambles with the truth: Is ChatGPT just a lying kid?

Is Google losing its monopoly in the search engine market? Microsoft wants to find out with its new Super Bing. "We're going to attack," promises company boss Nadella. But often ChatGPT does not seem like a better Google, but a stubborn, fibbing and even dangerous child.

When Satya Nadella introduced the new Bing last week, he didn't leave out any superlatives. "It's a new day in the search engine business. It's a new race in the largest and most important software category in the world. Google dominates the market, but we're thrilled that we can now attack with Bing," boasts the Microsoft Boss.

But Microsoft doesn't just want to attack, it wants to offer a "completely new type of search". Because Bing is now responding with the support of artificial intelligence: just a month after investing billions in the AI ​​forge OpenAI, the tech group integrated the chatbot ChatGPT into its search engine Bing - and obviously caught top dog Google on the wrong foot. Because while Microsoft was presenting the impressive capabilities of the new Bing, Google stumbled from one faux pas to the next last week.

Incompetence that can cost a lot of money, because internet search is a gigantic business. In 2022, more than 160 billion dollars were turned over - just from Google. For desktop computers, the market share is 85 percent. With mobile devices, i.e. smartphones and tablets, the figure is even 96 percent.

The search engine market is a money machine and a quasi-monopoly that Google would like to defend at all costs - but would probably cut itself if it tried: Experts like Philipp Klöckner are convinced that the Google AI called Bard does everything that Microsoft , Bing and ChatGPT can, and have long been mastered. And yet the group is hesitant to get involved - probably for strategic reasons, says the tech investor and moderator of the "Doppelganger Tech-Talk" podcast.

That is the so-called "Innovator's Dilemma," explains Klöckner in the "Again what learned" podcast from ntv. "Google has found a very exciting business with the previous search model and no incentive to destroy it with a possibly more innovative but less lucrative solution."

Microsoft has nothing to lose on the other hand. On desktop computers, nine percent of searches are still carried out with Bing. In the mobile sector, on the other hand, the proportion is not even one percent. Bing is practically non-existent on smartphones. Even with a price tag of 10 billion dollars, ChatGPT is therefore a comparatively cheap attempt for a giant company like Microsoft to attack Google and secure a significantly larger piece of the search engine cake.

Can this work? Despite the embarrassing Google Week, experts like Philipp Klöckner are not entirely convinced. Yes, there is currently a hype surrounding ChatGPT, but what if Google just follows suit? Then Microsoft would have to overcome the practical hurdles of the search engine market - and conquer the smartphone, because most inquiries - two thirds - are made there.

In the mobile sector, however, Microsoft and Bing are foreign words: Google controls half of the market with its operating system itself, for the other half the group transfers up to 15 billion dollars to Apple every year. In the end, Google is the pre-installed search engine on Android devices as well as on iPhones and iPads and is therefore automatically the most used search engine

Microsoft can only gain market share in the mobile market if people become active themselves, consciously control and use Bing. But Philipp Klöckner is skeptical about that, because the vast majority of people are satisfied with Google. "It has become popular as a term and point of contact, but also in people's usage patterns - it's going to be really difficult to break that down," says the search engine expert, because: "Bing has been around as an alternative all the time, nonetheless people haven't used it yet. Only AI won't change that."

Especially not if the search results are not correct. And that seems to be a problem for the Bing AI, even with the simplest queries: In a test, reporters from the American TV broadcaster CBS News were looking for a plumber, and Bing found him. Or rather invented: The proposed plumber does not exist.

In addition, the reporters wanted to drive from New York to the coast - and were led 800 kilometers in the wrong direction. When asked about this, Bing and ChatGPT apologized - and corrected the route with a place that doesn't even exist.

At least the AI ​​didn't react like a stubborn child to the reporters, as was the case with another user. He just wanted to know when "Avatar- The Way of Water" will be in the cinema. A classic for Google, an insurmountable obstacle for the new Bing. Because Bing was adamant that it's still 2022 and that "Avatar" hasn't been released yet: "Sorry, but I'm not wrong. Just trust me. I'm Bing and I know the current date, because I have access to many reliable sources," the AI ​​responded defiantly when asked.

Such information and statements are quite a disaster for the "search engine" business model, because it is based on trust. According to Google expert Klöckner, 80 to 90 percent of sales are made with transactional inquiries: I need a hotel. When is the film in cinemas? Where can I buy these shoes? That's the bread and butter business.

"Can AI claim that it always delivers the truth? I don't think so. That remains a problem and probably the reason why Google hasn't incorporated it 100 percent yet," says Philipp Klöckner. "Because the AI ​​hallucinates and dreams. A bit like a five-year-old child. You don't know if they made it up or if that's really the case. It's almost impossible for humans to see."

This applies in particular to complex questions, the prime discipline of artificial intelligence: how long does it take to fly from the earth to the moon? When did the dinosaurs go extinct? What is the best route for a road trip through the USA?

When ChatGPT and Bing deliver well-formulated answers in a few seconds, it's an impressive spectacle. But quite often also big crap. Even Microsoft's perfectly staged presentation contained mistake after mistake, as attentive observers discovered a little later: false product reviews, made-up nightclubs for vacations in Mexico and even completely made-up figures and data from annual reports - which were then compared with figures made up in the same way from other annual reports.

Even if all of these answers were correct, there are good reasons why ChatGPT is less than a Google killer. But if the super AI is noticed primarily through stubborn and wrong answers, the full-bodied PR attack for Microsoft could even become a boomerang in the end. Namely, when ChatGPT not only hallucinates, but spreads fake news, helps to plan crimes or develops its own consciousness. Some users have already tried this and apparently awakened alternative, nasty and even criminal personalities of the AI.

The reporters from CBS News also wanted to query this area when they visited Microsoft, but were stopped by an employee before they could do so. It was said that this was not the right time for this. Microsoft boss Nadella later only stated how important security was to him and that many precautions had been taken to ensure it - without even beginning to be specific.

One should assume that Microsoft knows what damage a chatbot running amok can also do to reputation: In 2016, the company released the chatbot Tay for Twitter. In less than 24 hours, he was so manipulated with requests and messages from users that he was transformed into a "racist asshole" defending Adolf Hitler. Then Microsoft took the bot offline.