The new Google AI should catch up with Microsoft

In the race for artificial intelligence, the company is upgrading its services with new functionality in the form of Google AI — while at the same time promising to take a prudent approach to do no harm. In addition to the search engine, other applications such as mail, office software and maps will be improved using artificial intelligence, CEO Sundar Pichai announced at the Google I/O developer conference.

One consequence of the cautious approach is that, for example, the company’s chatbot, Bard, controlled by Google’s artificial intelligence, will be available in German, but initially neither in Germany nor in the European Union. It is clear that the American company is trying to clarify whether Bard is compatible with the legal framework in the European Union.

Google wants to experiment with important innovations for its most important product – Internet search. With the help of artificial intelligence, the search engine should be able to answer the question, for example, which two nature parks are more suitable for a family with children and a dog. The answers are formulated in complete sentences, in addition to the usual internet links. With Google AI, importance must be placed on the ability to display sources of information – something that is unnatural in other AI applications.

When searching for a bike, you are shown suitable offers from dealers as well as advice on things like suspension. In addition, one can, for example, get advice on hand signal rules when cycling in California, said Google director Cathy Edwards. “These are things you would never have asked for in a search before.”

Research in artificial intelligence is initially limited to testing. The program decides for itself when to answer the search query in this way. They say they want to better understand the cases in which this makes sense.

Google still does most of its business through ads in the Internet’s search environment. Most advertisers pay to have the search query links appear at the top of the screen. So far, an open question is what impact the prevalence of AI-based detailed answers will have on this business model.

Pichai also showed how the software can craft a speech with just a few specifications. If you write a story, the program should be able to offer suggestions for more story twists and auto-generated illustrations. In the Google Photos app, you will soon be able to not only remove unwanted objects and people, but also, for example, change your position in a photo. Missing details that originally remained behind the edge of the screen have to be added by the program itself.

Google has been working on AI-based applications for years, but is currently under pressure to reveal more of them. At the end of last year, startup OpenAI kicked off a new AI competition when it made its ChatGPT chatbot public. The program has caused a stir because it can form sentences like a human. They are trained on massive amounts of data and estimate word for word how a sentence might go. This carries the risk that it can give out completely false information.

Google’s archenemy Microsoft has a multi-billion dollar agreement with OpenAI and is bringing AI software into its applications across the board. So far, Google has been reluctant to do so, citing responsible use of the technology.

At the Google I/O conference, the group stuck with it. “The only way to be brave in the long run is to act responsibly from the start,” said James Manica, Google’s social responsibility officer for the use of artificial intelligence. The group sees the risk that the software could reinforce biases or be used to produce false information. To protect against this, files created with the help of Google AI must be provided with metadata so that they can be instantly recognizable.

Manica said Google will also provide only certified developers with software that can automatically create dubbed versions of videos. This is to prevent so-called deepfakes with the alleged actions of real people. At the same time, Manica emphasized that Google decided years ago not to make interfaces for facial recognition apps available to the public.

For the new capabilities of Google AI, the company introduced a new language model called Palm 2 to compete against OpenAI’s GPT-4. Palm 2 can master more than 100 languages ​​and brings writing, programming and analysis skills with it. The Google Bard chatbot will also be based on the Palm 2. Google AI answers to chat questions, which can only be experienced in the US and UK so far, will soon be available in 180 countries in English, Korean and Japanese. However, in European Union countries, Bard will not be available at the moment, including Germany. However, support for German and 39 other languages ​​will follow soon.

At the developer conference, Google also introduced three devices. On the other hand, the group introduced the Pixel Fold smartphone, which folds to the size of a small tablet and competes with the Samsung Galaxy Fold and similar devices from Chinese manufacturers. Google also showed off the Pixel 7a, which is a slightly watered down version of the previous top model, the Pixel 7 Pro.

The Pixel Tablet was also introduced. A special feature of the Google tablet is an additional stand with a speaker that can be used as a constantly active desktop device. While the Pixel 7a is now available for €509, those interested in the foldable smartphone and tablet will have to wait.