Chatbots and AI: What can ChatGPT do?

Write emails, applications and articles about Goethe – chatbot ChatGPT does all this. The company behind it could soon become one of the most valuable startups in the world. But there is also a lot of criticism.

By Antonia Manweiler, tagesschau.de

In science fiction movies, an artificial intelligence that can have normal conversations with humans is no longer a great invention. It is part of everyday life. But experts assume that this scenario is still some time away.

However, since the end of November 2022, users of the ChatGPT chatbot have been able to get an experience that at least goes in this direction: the computer program can answer questions on a variety of topics, such as how far the Sun is from Jupiter or why Johann Wolfgang von Goethe is one of the most important German language poets. On demand, the dialogue system synthesizes its scripts in a more humorous manner:

ChatGPT chatbot can translate texts, write scripts, applications, emails, complete articles or computer codes. The acronym “GPT” stands for “Generative Pre-training Transformer,” because the chatbot learned human-like communication through countless forays into the Internet and reading numerous texts.

Elon Musk and Peter Thiel as financiers

The OpenAI artificial intelligence research lab from California is behind the development of the chatbot. Its foundation was funded in 2015 by notable investors from Silicon Valley, such as Tesla boss Elon Musk, technology investor Peter Thiel, and LinkedIn co-founder Reid Hoffman. Sam Altman, who now runs the company, was also one of the investors who gave the company $1 billion to get the project started.

OpenAI was founded with the goal of advancing digital intelligence. Another idea was to have a groundbreaking research facility once human-scale artificial intelligence was within reach.

OpenAI originally intended to be a non-profit organization, and abandoned this status four years later – in order to have better access to capital. Some accuse the company of throwing its ideals into the sea.

OpenAI has moved away from the original goal of creating value for everyone, not shareholders. Only shortly after the nonprofit ended, Microsoft paid the company $1 billion in 2020 to acquire the exclusive license to OpenAI technology. In partnering for technical possibilities, “most of which we can’t even imagine,” Microsoft wrote at the time.

Potential billion dollar deal with Microsoft

Now Microsoft can expand that partnership even further with a $1 billion deal. This was recently reported by the American news portal “Semaphore”. Thus, the possibility of Microsoft’s $10 billion stake is under discussion. This will increase the value of the AI ​​company to $29 billion, making OpenAI one of the most valuable startups in the world. According to Semaphore, the group must receive 75 percent of all profits from OpenAI in order for the initial investment from Microsoft to be recouped. According to this, Microsoft could own almost half of the company at 49 percent.

Currently, OpenAI’s business costs a lot of money. Sam Altman, co-founder and head of OpenAI, wrote on Twitter that the company pays a few cents for computing power every time the chatbot is used. The company was said to have spoken to investors about the expected income of $200 million for 2023, according to the Reuters news agency, that it expects next year. Even with revenues of a billion dollars. However, it is unclear how much this covers costs.

Soon part of the search engine?

According to technology portal ‘The Information’, Microsoft is said to be working on a new version of ‘Bing’ search engine. Apparently, the idea is that ChatGPT technology should be used to compete with the Google search engine. In any case, Microsoft can use collaboration to break into the field of artificial intelligence, which Google subsidiary Alphabet is also pursuing. The tech giant is also said to be considering integrating OpenAI functionality into programs like Outlook or Word.

Elon Musk withdrew from the company in 2018 to avoid potential conflicts of interest with Tesla, the electric vehicle maker he runs, which also deals with artificial intelligence. Since then, Musk has repeatedly criticized OpenAI, for example over a lack of transparency or the end of non-profit status.

New York School Block website

There is also criticism of the chatbot, which is currently only in beta, from others. Just a week after ChatGPT was released, 1 million users have already tried the platform, including many schoolchildren and students, as word spread that the chatbot writes essays and homework assignments in seconds. There was already a reaction to this in New York. There, access to the website is now blocked in public schools. Other schools and universities around the world are also discussing how to use chatbots.

Claims that ChatGPT is simply inventing false writing things or contributing to the massive spread of false information are particularly impactful. If you ask the chatbot itself if it can answer every question, it says no.

In the end, a chatbot is only as reliable as the data it is fed into. So the software is up to date with dates of 2021. In an earlier version, GPT-2, the chatbot was so good at creating fake messages that OpenAI decided not to publish.