How should AI companies approach the topic of safety as they develop new technologies and products?

"We must force the AI to make concessions based on our own limitations (people aren't all good, efficient, etc..). This needs to be a tool for humanity, not something better than humanity with free reign. It needs to fit into our society to change it slowly, with small progressive steps."

• Regulation of AI should be as light as possible. The only purpose of such regulation should be to: 1) Encourage rapid development, adoption and innovation; and 2) Discourage bad actors - companies, people or nations who would use this technology to harm others. That's it.'

"I think the only sensible regulation will be transparency rules - clearly show people where they interact with an AI, what content was created by an AI, and possibly what content was used to train the AI."

• "There will be no actual regulation. The genie is out of the bottle. Any real regulation would have to be global in scope. It’s a nice thing to talk about. Makes it look like governments are paying attention. In reality, it’s an impossible task."

“It’s hard for me to care about any potential dangers posed by AI when climate change poses a much more real and existential threat to humanity.”

• “The real threat at this point is almost 0. We will see potential risks as they are exposed and deal with them accordingly. This is a manageable technology that will prove to be very disruptive in a good way. Our greatest risk is slowing the advancements down and not allowing AI to save lives in medicine and other critical areas.

“Young people are excited about AI's possibilities, but they need to have the critical thinking skills to evaluate what AI produces. I am not in the "ban AI" camp, but I think safety is critical, and we need to be protecting students from and teaching them about AI's fallibilities.”