"Super intelligence"

ChatGPT to receive improved thinking ability

ChatGPT is apparently getting closer to its goal of "superintelligence". According to insiders, developer OpenAI wants to equip its artificial intelligence with an improved thinking ability developed under the code name "Strawberry" within the next two weeks.

This was reported by the industry service "The Information" on Tuesday, citing people familiar with the matter. However, it is unclear how the new technology, which initially only generates texts, will be marketed. Neither OpenAI nor its partner Microsoft were initially available for comment.

"Strawberry" is intended to give ChatGPT the ability not only to provide answers to queries, but also to think ahead, for example to independently search the internet for information. This is where the current large language models (LLMs), which form the basis of AI, fall short.

Concerns about superior AI
At the end of last year, some OpenAI developers mentioned a breakthrough on the way to an artificial "superintelligence" in an incendiary letter and warned against making an "Artificial General Intelligence" (AGI) superior to humans available to the public without sufficient risk assessment.

AGIs can handle complex tasks without outside interference. Critics fear that these programs could then control defence systems, spread political propaganda or produce weapons.