Google’s AI presentation was a disaster

ChatGPT will be an important part of Bing in the future. While Microsoft’s search engine has been smiled at so far, the tide could turn with it. In a few weeks you could have real added value compared to Google.

But the top dog does not sleep. Google is also planning to make its own language model called LaMDA available to the public, at least in a limited form. A text AI called Bard is now being tested in a limited circle for this purpose.

On February 6th, Google officially introduced Bard. And showed in a demonstration what the chatbot should be able to do in the future. The problem: The AI’s answer contained a false statement. And that cost Google dearly on the stock exchange.

link to Twitter content

As part of the demonstration, Google is showing a short video showing Bard in action. It asks the AI ​​the following question:

Bard then responds with three examples: a series of galaxies called Green Peas, images of galaxies older than 13 billion years, and the first image of a planet outside our solar system.

However, the third example was a false statement. Because the corresponding image was already taken in 2004 with the VLT (Very Large Telescope) in the Atacama Desert.

That AIs regularly make mistakes of this kind is nothing new. Just recently, the US site CNET had to correct a large number of articles that were co-authored by an AI.

39
5

more on the subject

One of the biggest tech sites has an AI write articles

What is unusual, however, is that the AI ​​error made it into Google’s demo video without being discovered beforehand.

But that’s not the only blunder that Google has made during the presentation of new AI tools in recent days. While of the Live from Paris

-Events the intention was actually to present the new multi-search function to the public – but the necessary smartphone could not be found and the demonstration had to be cancelled.

The first error in particular caused Google’s share price to plummet on the stock exchange. Within a few hours, it lost almost 8 percent of its total value – a total of more than $100 billion.

Editor’s opinion

Alana Friedrichs: Sure, AIs make mistakes. This is nothing new. And that something can go wrong during a presentation is only human. The combination of both incidents does not put Google’s actions in the best light. The search engine provider seems to act hastily and does not take the necessary time for quality control.

Of course, it’s important not to lose touch. If Google wants to maintain its supremacy, one has to find an answer on Bing and ChatGPT. But Google needs to move away from the notion of being first in the race for AI capabilities. The train has already left.

Instead, it is now about being better than the competition. Something I have no doubts about Google. It was not for nothing that Google developer Blake Lemoine made headlines last year when he attested to the awareness of the LaMDA language model.

However, mistakes like the above two mean that Google could lose its earned reputation as a leading AI company. Mistakes of this kind are exactly what you can’t afford if you want to convince later with quality and want to beat more established models like OpenAI’s GPT-4 in the eyes of the public.

36
2

more on the subject

Instructions for ChatGPT: How to use the text AI for free in Europe

What do you all mean? Does Google now want to rush out its own text AI and is stumbling? Is the company actually on the right track and just a victim of unfavorable coincidences? Or is Google’s golden age over with the advent of ChatGPT? Let us know what you think about the current situation in the comments!