By: Nick Gambino
This is the year of AI. Everyone’s talking about it and almost every major tech company is launching their AI bot. With all the wonder that it brings, there’s a dark side to AI. No, I’m not talking about Terminator or Matrix scenarios but abuses that can occur when humans rely too heavily on an unproven new tech.
President Biden signed an executive order on Monday, putting in place safeguards against the unfettered use of AI while also embracing it in a meaningful way.
“AI is all around us,” Biden said at the signing. “To realize the promise of AI and avoid the risk, we need to govern this technology.”
The new executive order lays out several guidelines and requirements to help ensure this. Calling up the Defense Production Act, it requires that anyone developing new AI models that potentially pose safety risks to the country or individuals share with the government all safety test results. It essentially demands transparency so that nothing potentially dangerous in the AI space is developed without the proper guardrails in place.
The order also instructs the creation of standards and testing to ensure things don’t get out of control. As part of that, Homeland Security is going to establish an AI Safety and Security Board to oversee it.
Another instruction that should help keep AI in check is for the Commerce Department to issue guidelines for watermarking all AI creation so as to distinguish it from real creation by humans. This goes for AI-created videos like Deepfakes or documents or articles written by an AI bot.
It’s not only about handcuffing AI, but steering it in the right direction so as to not let it get out of control. Biden added funding for research into AI and had a new website set up, AI.gov, to connect people with government jobs in the field of emerging AI.
There’s a lot to cover in the order. You can read President Biden’s entire executive order on AI here.