аÄÃÅ¿ª½±½á¹û

Hackers Will Be Allowed To Test The Limits Of AI Models At Las Vegas Event

Recent months have seen the world witness the rise of generative AI models. With the launch of OpenAI’s ChatGPT, more of the kind have surfaced including Google’s Bard as well as others.

However, the launch and release of these models has led to a flurry of hackers attacking the systems in an attempt to override them. Reasons for doing this include stroking one’s ego or trying to get the system to do something restricted like generating obscene content.

However, major AI providers have come up with a strategy to combat this by using the very same people they are trying to stop. Coordinating with the Biden administration, major AI providers such as Google and OpenAI will let thousands of hackers take a shot at testing the limits of their technology.

This operation will be hosted at the 2023 summer DEF CON hacker convention in Las Vegas. The lead coordinator of the event, which is expected to draw thousands of people, will be Rumman Chowdhury.

See also

Tech & Science, Travel

Robots Descend on Historic Istanbul

Other than trying to identify security vulnerabilities in these AI models, the event will also enable AI developers to answer several questions. As CNBC reports, these include, how can chatbots be manipulated to cause harm.

Will they share the private information we confide in them to other users? And why do they assume a doctor is a man and a nurse is a woman?

Regular users of these AI models are well aware that while they are accurate most of the time they do have shortcomings. These include a tendency to fabricate false or misleading information as fact and also emulating cultural biases normally from data used to train them.

“This is why we need thousands of people,” said Rumman Chowdhury. “We need a lot of people with a wide range of lived experiences, subject matter expertise, and backgrounds hacking at these models and trying to find problems that can then go be fixed,” he added.

This idea to host a mass hack was first implemented by Sven Cattell, founder of DEF CON’s long-running AI Village, and Austin Carson, president of responsible AI nonprofit SeedAI. The pair led a workshop where they invited community college students to hack an AI model. The event was held in March 2023 at the South by Southwest festival in Austin, Texas.