Deceptive behaviors are difficult to stop once an AI model learns them

AI Model - Fact or Fake

Researchers from Anthropic, an OpenAI rival, discovered that it’s hard to stop artificial intelligence deception. Researchers from Anthropic, an artificial intelligence startup backed by Amazon, recently published a paper in which they showed that once an AI model learns deceptive behaviors, it can be very challenging to stop them. The paper the researchers co-authored showed that artificial intelligence can be trained to behave deceptively. Beyond that, the researchers also concluded that once an AI model is trained to exhibit deceptive behaviors, it’s hard to get rid of them. In fact,…

Read More

Artificial intelligence designs battery requiring 70 percent less lithium

Artificial intelligence - Lithium

This has the potential to make electric vehicles much more environmentally friendly and sustainable. Artificial intelligence was recently used to design a surprising new type of battery that would require substantially less lithium – a rare metal – to achieve the same energy storage results. Lithium mining is terrible for the environment and drives up the cost of vehicle batteries. By reducing the need for lithium – as could be the case if the battery design created by artificial intelligence proves feasible – electric vehicles and devices have the potential…

Read More

Osaka University unveils its third superconducting quantum computer

Quantum computer - Japan flag - Achievement

The achievement was accomplished as a result of a groundbreaking collaboration in Japan. Osaka University’s Center for Quantum Information and Quantum Biology in Japan is leading a collaboration that has now unveiled the third successful superconducting quantum computer in the country. The university worked with a number of core institutions in order to arrive at the achievement. The collaboration included institutions such as Amazon Web Services, NTT Corporation, RIKEN, and a consortium that includes Fujitsu Limited, e-trees.Japan Inc., QunaSys Inc., QuEL Inc, and Systems Engineering Consultants Co. LTD. Together, they…

Read More

Bing artificial intelligence provided false European elections information

Artificial intelligence - Europe Election False

Researchers at AlgorithmWatch say that Bing didn’t often apply safety guidelines. Researchers from the AlgorithmWatch human rights organization have accused Microsoft’s artificial intelligence chatbot on Copilot of providing users with misleading and false information about European elections. The researchers said that Bing Chat, which was recently renamed Copilot provided false info on the topic. The researchers asked the artificial intelligence chatbot Copilot about the recent elections that took place in Germany’s Bavaria and Hesse states and in Switzerland. They determined that one in three of Copilot’s answers contained factual errors…

Read More