Where does all the hype about ChatGPT come from?

"It relies mainly on its ability to generate very natural and readable speech, which makes it very useful for many applications," answers Open AI's new chatbot.

Patrick Schlereth

Editor on duty at FAZ.NET.

  • Follow I follow

The machine can't know anything about the hype that its release caused at the end of last year.

Your knowledge base will last until 2021, but you have not yet been connected to the Internet.

But it is said to have the potential to revolutionize the Internet.

Some even speak of the end of Google's dominant market position.

The idea: Instead of a long list of results, users get exactly the right answer - and in normal everyday language.

Is ChatGPT Smart?

ChatGPT can not only translate texts, but also write applications, screenplays or poems in impressive quality.

The abbreviation "GPT" stands for "Generative Pretrained Transformer", the individual words describe how the AI ​​is constructed: The algorithm has appropriated the human-like communication through numerous data from the Internet.

His answers are based on the statistical probability that a given word almost always follows the other.

Whether the artificial intelligence behind it can actually be described as intelligent depends on the definition of intelligence.

The machine cannot think or develop consciousness, but it can link countless data in a matter of seconds.

"Language is the most interesting manifestation of intelligence," says German-born AI specialist Richard Socher at the DLD tech conference in Munich.

Measuring the progress of artificial intelligence in terms of human intelligence is already limiting for the debate.

"A search engine can do much more than a librarian."

Open AI boss Sam Altman himself has admitted that the answers from ChatGPT are error-prone.

"ChatGPT is incredibly limited, but good enough at some things to create a misleading sense of awesomeness," he tweeted in December.

In important questions one should not rely on the chatbot.

Half of Silicon Valley has invested

Nevertheless, ChatGPT is already being measured by its potential.

At the DLD, the chatbot is one of the dominant topics, the majority of the verdicts are negative.

ChatGPT's answers are "plausible, but decoupled from reality," says Phil Libin, founder of the note-taking app Evernote and current head of the AI ​​studio All Turtles.

The criticism is primarily directed at the lack of transparency, because the questioner does not find out from which sources the chatbot compiles its answers.

You only realize that the answer is wrong if you already knew it beforehand.

At this stage, Jonas Andrulis says ChatGPT is best suited to having “Jar Jar Binks-style declarations of independence written.”

When Open AI was launched in 2015, the founders had set themselves noble goals.

The non-profit organization wanted to counterbalance the commercial development of artificial intelligence in the leading corporations.

"Our goal is to be valuable to everyone, not just shareholders," damlas said in a blog post.

Silicon Valley liked the idea: the investors included tech investor Peter Thiel, Linkedin founder Reid Hoffman and Elon Musk, who has been warning of the dangers of artificial intelligence for mankind for years.

The Tesla boss left the Open AI board of directors in 2019, and in the same year Microsoft invested billions.

With the development of ChatGPT, the formerly non-profit organization could now become one of the most valuable start-ups around.

According to media reports, Microsoft wants to invest another ten billion dollars in Open AI and integrate the chatbot into its Office applications.

The Microsoft search engine Bing in particular could benefit from ChatGPT and re-establish itself as a serious competitor.

Google has allegedly already proclaimed the "Code Red" and is considering making the language intelligence LaMDA, which has only been used internally so far, publicly accessible.

For Musk, Microsoft's entry into Open AI is "the opposite of open," as he wrote on Twitter in 2020.

Two years later, he describes ChatGPT's capabilities as "frighteningly good" and thinks humanity is not far from "dangerously powerful AI".

Evernote founder Libin doesn't see the future as bleak, in any case the "end of the world" is not imminent.

Certain technical habits would be overtaken by current developments.

For example, ChatGPT could compose emails for users in the future.

"But we hate email anyway."