In the primaries, I learn almost everything I know about third party candidates from the Voters Pamphlet. If they sound intriguing, I try to look them up online, usually with limited success.
I was under the impression that microtargeting was documented in the 2016 election-that Cambridge Analytics has done it using data from Facebook on behalf of Trump.
Cambridge Analytics didn't work with him once he won the primary and had access to RNC's data. The internal data of both parties is significantly better. Also, Facebook has been selling data for years and years. Why do people think this started politically with Cambridge Analytica?
As a software developer who’s dealt with many computer security issues, I really like the idea of using digital signatures on news media. The main issue I can see with it is that while it can work well for original photos, videos, and _plain text_ documents, it can get messy when you need to integrate these things into Web pages where lots of other formatting, fluff, and even overlays are inserted alongside or on top of the media that people need to verify.
Can your interviewee cite the sources of his claim that campaigns aren't still micro targeting. I suspect the trump campaign has relied heavily on micro targeting in 2016 and 2020 with data sets hoovered up via the fb Cambridge Analytica scandal
New technology always seems so realistic at first. A few years later, we recognize it easily. Study it, pay attention, learn what to listen or look for, what to be suspicious of.
Digital signatures are useful, but digital timestamping (described on the Wikipedia article "Trusted timestamping") is a way to show that a document was produced in a narrow window of time. Presumably this makes it impractical to make a fake in that time.
The best part about FiveThirtyEight these days are the expert guests. They really do know their stuff, and somehow, that is surprisingly rare in this media environment. Plus, Galen is good at pushing the interviewees (if sometimes a little overboard).
I think that with new generative ai, we are at the stage of identifying problems. I think answers will result from discussions and opinions from many different people. My hope is that this guy wants people to talk about the questions. What do you think about this idea?
Bueno de Mesquita's remark about A/B testing rings a little underdeveloped. Bots' ability to iterate and test at scale is so much more than what humans can do that it is an entirely different beast. I'm hoping I'm wrong in this case and the factors of time delays in the meatspace world put a massive penalty on messaging optimization timeframes.
The real power of AI isn't in the flashy front-end, but in the unsexy background. Think improved data analytics, the stuff that Peter Thiel is doing. Or what the military is doing to try to integrate it. There are so many AI models that we don't know about, but making steady progress anyways.
One can only hope that the percentage of people who don't know what deep fakes are decreases to zero and that the percentage of those who are able to identify something as deep fake increases.
Maybe the data embedded into an image of where that image came from can be stored in a blockchain. Thus finding a useful nonscammy productive use of the technology to offer transparency about what images are photos, which have been manipulated, which are completely fabricated, etc
I am liking the 538 YT pieces. Wondering, is there a universe in which Dark Galen feeds some prompts into. ChatGPT and takes over the world! Here is a start: make a user friendly button in which a potential voter can ask, What candidates are my contact list likely to vote for in 2024?
aI is like really smart. I don’t see why we don’t give it some ability to see if the public elected the right person. We might want a candidate but the AI might be smart enough to say wait a min, you humans choose wrong and maybe have some kinda override in case people say elect trump again or someone like him.
You're not very smart if you think AI right now actually has some sort of thought of its own. It's just an amalgamation of collected data. Input more left wing data and arguments? It will have more left leaning positions. Input more right wing data and arguments? It will lean right.
@@CanadianEhHole I know it does not have thought. But if we let people like bill gates, Obama, or any other geniuses program it then it will be more inclined to help humanity even if the public is to stupid to understand. For example if we had some kinda AI oversight and in 2024. The public votes trump back into office. The AI would know based on data that it is bad for the society, economy, humanity and the earth as a whole and could veto, allowing for a correction of the infraction
In the primaries, I learn almost everything I know about third party candidates from the Voters Pamphlet. If they sound intriguing, I try to look them up online, usually with limited success.
Tuning into third -tier- party candidates livestreams with another.. dozen people. At least they'll answer your questions 😸
I was under the impression that microtargeting was documented in the 2016 election-that Cambridge Analytics has done it using data from Facebook on behalf of Trump.
Exactly what I was thinking. 2016 was “the first” A.I. Election.
Cambridge Analytics didn't work with him once he won the primary and had access to RNC's data. The internal data of both parties is significantly better.
Also, Facebook has been selling data for years and years. Why do people think this started politically with Cambridge Analytica?
As a software developer who’s dealt with many computer security issues, I really like the idea of using digital signatures on news media. The main issue I can see with it is that while it can work well for original photos, videos, and _plain text_ documents, it can get messy when you need to integrate these things into Web pages where lots of other formatting, fluff, and even overlays are inserted alongside or on top of the media that people need to verify.
Can your interviewee cite the sources of his claim that campaigns aren't still micro targeting. I suspect the trump campaign has relied heavily on micro targeting in 2016 and 2020 with data sets hoovered up via the fb Cambridge Analytica scandal
The RNC and DNC have better data sets than Facebook. It's why Cambridge Analytica was not used post-RNC primary victory by Trump.
Really great podcast, thanks!
New technology always seems so realistic at first. A few years later, we recognize it easily. Study it, pay attention, learn what to listen or look for, what to be suspicious of.
the thing here to fear is that the AI will do the same and do it better and faster, what do you think about that ?
I might be a little stoned but this conversation is freaking me the fuck out. I don’t want to be a paperclip!
Digital signatures are useful, but digital timestamping (described on the Wikipedia article "Trusted timestamping") is a way to show that a document was produced in a narrow window of time. Presumably this makes it impractical to make a fake in that time.
It means we are completely, absolutely and irreversibly fucked....
The best part about FiveThirtyEight these days are the expert guests. They really do know their stuff, and somehow, that is surprisingly rare in this media environment. Plus, Galen is good at pushing the interviewees (if sometimes a little overboard).
The gentleman being interviewed is better at identifying problems than solutions.
I think that with new generative ai, we are at the stage of identifying problems. I think answers will result from discussions and opinions from many different people. My hope is that this guy wants people to talk about the questions. What do you think about this idea?
Couldn't citizens be framed with AI composing false crime scenarios? If the institutions are corrupted?
Bueno de Mesquita's remark about A/B testing rings a little underdeveloped. Bots' ability to iterate and test at scale is so much more than what humans can do that it is an entirely different beast. I'm hoping I'm wrong in this case and the factors of time delays in the meatspace world put a massive penalty on messaging optimization timeframes.
Personally, I don't use chatbots for customer service, so why would I use that for politics? The FAQ, or Google, is quicker.
The real power of AI isn't in the flashy front-end, but in the unsexy background. Think improved data analytics, the stuff that Peter Thiel is doing. Or what the military is doing to try to integrate it. There are so many AI models that we don't know about, but making steady progress anyways.
One can only hope that the percentage of people who don't know what deep fakes are decreases to zero and that the percentage of those who are able to identify something as deep fake increases.
Maybe the data embedded into an image of where that image came from can be stored in a blockchain. Thus finding a useful nonscammy productive use of the technology to offer transparency about what images are photos, which have been manipulated, which are completely fabricated, etc
I didnt know robert downey jr was getting into ai and politics
*AI Selection
It's all a lil scary...
I am liking the 538 YT pieces. Wondering, is there a universe in which Dark Galen feeds some prompts into. ChatGPT and takes over the world!
Here is a start: make a user friendly button in which a potential voter can ask, What candidates are my contact list likely to vote for in 2024?
aI is like really smart. I don’t see why we don’t give it some ability to see if the public elected the right person. We might want a candidate but the AI might be smart enough to say wait a min, you humans choose wrong and maybe have some kinda override in case people say elect trump again or someone like him.
You're not very smart if you think AI right now actually has some sort of thought of its own. It's just an amalgamation of collected data. Input more left wing data and arguments? It will have more left leaning positions. Input more right wing data and arguments? It will lean right.
@@CanadianEhHole I know it does not have thought. But if we let people like bill gates, Obama, or any other geniuses program it then it will be more inclined to help humanity even if the public is to stupid to understand. For example if we had some kinda AI oversight and in 2024. The public votes trump back into office. The AI would know based on data that it is bad for the society, economy, humanity and the earth as a whole and could veto, allowing for a correction of the infraction
You know that these right wingers are going to be abusing this!
Are deep fakes *really* all that different from Photoshop? Sounds like feigning guilt to claim credit.
Literally laughing out loud when he basically says an AI can give better speeches than Ron DeSantis. 😂