Like it looks at our interests rather than doing something quite wacky and being very biased.
But I think since generative AI has come about, a lot of people have been looking into it from the point of view of, okay, are these large language models or generative AI, are they aligned with our values or are they acting in a way how we act or what we expect? from each other.
So like that's been one of the biggest topics in that space, like make AIs go well in a way so that we don't constantly have to tune it for it to work for us because it is pretty uncontrollable at the moment.
There's just no way we can always catch up given the progress that we have seen over the last four years, even five years.
So safety really is not just responsible AI, but to make AI work for us in a way that it aligns with how we see the world how we act with the world it doesn't like cheat or it doesn't work in a way that's kind of just being duplicitous or making us confused at all making us feel like you know it seems aligned with how we would expect it to work but yeah Eventually you find out it's not.
There is no like set definition other than just making sure that the AIs you currently have, they are safe enough for us to be using or deploying or even yeah.
Like it looks at our interests rather than doing something quite wacky and being very biased.
But I think since generative AI has come about, a lot of people have been looking into it from the point of view of, okay, are these large language models or generative AI, are they aligned with our values or are they acting in a way how we act or what we expect? from each other.
So like that's been one of the biggest topics in that space, like make AIs go well in a way so that we don't constantly have to tune it for it to work for us because it is pretty uncontrollable at the moment.
There's just no way we can always catch up given the progress that we have seen over the last four years, even five years.
So safety really is not just responsible AI, but to make AI work for us in a way that it aligns with how we see the world how we act with the world it doesn't like cheat or it doesn't work in a way that's kind of just being duplicitous or making us confused at all making us feel like you know it seems aligned with how we would expect it to work but yeah Eventually you find out it's not.
There is no like set definition other than just making sure that the AIs you currently have, they are safe enough for us to be using or deploying or even yeah.
The rest of this transcript — segmented and speaker-labeled, so you land on the exact moment something was said
Search every transcript — by keyword, by phrase, or by meaning, across every show Radar indexes
Trends — what is surging across podcasts, measured against its own baseline
Alerts — when a name you follow appears in a newly indexed episode
No account is needed to search Radar.