Submind YouTube summaries
Thumbnail for “AI Is Being Used To Rewrite History” - Former CIA Agent Reacts

“AI Is Being Used To Rewrite History” - Former CIA Agent Reacts

Watch on YouTube

Video summary

The podcast opens with a discussion on how artificial intelligence (AI) and digital editing are fundamentally altering our perception of reality, using Alicia Keys' performance at the Super Bowl as a prime example. The speaker notes that while her voice initially cracked during a live piano solo, post-event recordings available online show a perfectly tuned vocal track, suggesting retroactive fixes were applied to erase this historical moment from public memory. This phenomenon extends beyond mere autotuning; it represents a dangerous capacity for AI to misrepresent history by erasing the present and rewriting recorded events. The speaker expresses concern that as these technologies become more sophisticated, distinguishing between authentic recordings and manipulated media becomes nearly impossible without specialized assistance, thereby threatening the integrity of historical records. This erosion of truth is compounded by shifts in journalistic standards and the rise of generative AI tools like ChatGPT and large language models (LLMs). Historically, credibility relied on named sources and rigorous verification processes; however, modern outlets now frequently publish stories based on anonymous information, making it difficult for individuals to verify clips seen on platforms like X or YouTube. The speaker argues that while the goal of search engines is to reduce friction in finding answers, this convenience often comes at the cost of accuracy. When AI provides quick but potentially incorrect summaries, users are less inclined to conduct their own research because the path of least resistance favors accepting convenient information over verifying facts through time-consuming investigation. The conversation then shifts to the mechanics of deception and why telling the truth remains inherently easier than lying, a principle emphasized during intelligence operations in the field. The speaker recounts advice from former bosses who encouraged admitting mistakes immediately rather than covering them up, noting that punishing individuals for errors only disincentivizes honesty. In an interrogation context, liars must expend significant mental energy maintaining consistent falsehoods and remembering conflicting details they have invented, whereas truth-tellers can simply recount events without fear of contradiction. However, the speaker warns that technology is shrinking this gap; AI now allows propagandists to generate disinformation campaigns with unprecedented speed and efficiency compared to old methods like cultivating networks of sympathetic journalists or creating memes manually. The scale of modern misinformation has evolved from labor-intensive operations involving human creators into highly automated systems capable of targeting specific demographics across social media platforms like Facebook, Instagram, and TikTok. The speaker references the Cambridge Analytica scandal as a precursor where targeted ads were used to sway voters by exploiting psychological vulnerabilities, but notes that current AI capabilities allow for the creation of thousands of fictional personas with followers who amplify fake narratives automatically. This acceleration means that state actors such as Russian or Chinese intelligence agencies can now deploy elaborate disinformation campaigns much more effectively than in previous decades, potentially influencing upcoming elections globally. Ultimately, the podcast concludes that while we cannot eliminate human nature's tendency to consume information passively and follow paths of least resistance, society must recognize that this problem is accelerating rapidly. The ability for bad actors to create deepfakes or bot-generated content has reached a point where it will inevitably impact future political landscapes worldwide, not just in the US. As AI continues to dominate content creation—potentially making human-authored posts rare within five years—the burden falls on individuals and institutions to navigate an environment where verifying truth requires more effort than ever before. The speaker suggests that without significant changes in how we approach information consumption and verification, the integrity of our shared reality will continue to be compromised by these powerful technological tools.
Read the full video transcript
did you see what happened with the Super Bowl where Alicia Key's voice broke and then they went back and retroactively fixed her voice do you see this no I I you know I did not the Giants were not in the Super Bowl so I didn't watch it right well look you missed you missed Alicia Keys very good performance Usher fantastic dancer still pretty ripped um Alicia ke's voice broke on the first note of this piano solo thing that she was doing but if you go back and watch the Super Bowl's recording of it on YouTube her voice is perfect W and this has been I look Alicia Keys is not the Vanguard of us rewriting history but you can certainly see between AI being able to misrepresent how history would have been this sort of erasing of history and rewriting of the present and then this retroactive changing and even if it's just a little bit with regards to someone's voice this is different to autotuning a song this is a record of what happened in history right and and I think I I'm this is starting to get quite worrying like I ER on the side of non-c conspiracism but yeah this is starting to get a bit I don't like the doors that are being opened to you well and I think there's there are some uh mechanisms that are and companies that are trying to uh or have been trying to get outad of the curve I'm not sure that that's that's possible right now but um you have to it's it's like it's like the old days you you'd read an article in newspaper and you think okay that's very interesting well you know but is it true right I mean so what's the basis for this you know and used to be one of the defenses was sources would be named right you couldn't write an article you you wouldn't get it published if it was based on Anonymous sources well now that's not true anymore New York Times will publish a front page article on the you know top half based on Anonymous sources and so you're trying to verify something right and and you push that forward to today and you say okay well now I'm trying to verify um a a clip that I'm seeing on on X right or on YouTube or something and how do you do that because the ability to manipulate video abil and audio is remarkable right there's there's you almost impossible to tell the difference basically without some assistance and so there's companies out there that are trying to ensure that you are basically imprinting actual the the truth right so you take video and that is the the original source and it's it's imprinted in the in the code that says this is the the date the time the location and so you can verify that what's being said by that political candidate or whomever right on that in on that Podium is accurate because then you may see a day later you may see the same clip but it's been altered aot tuned and yeah autotuned and so there are some companies out there that are working to do that to to prevent deep fakes but it's it's very difficult because you're relying on people's curiosity and um and them having the interest right to pursue to make sure that what they're seeing and most people don't either they don't have the time or they don't have the patience or the interest well the point with chat GPT and these llms and Ai and search engines is to reduce the friction from getting where you are to where you want to be in terms of understanding in terms of whatever it is that you're searching for right and the fact that it's friction is precisely where the attraction lies it's the convenience and all the rest of stuff right and if you could imagine a world in which the most convenient route also bakes in uh lack of accuracy so it goes look you can find this out in 3 seconds but there's a a accuracy yes EXA precisely you could be a little bit wrong either way or you can do your own research it's going to take you a day and a half and you're going to know what actually happened where were people to do of course of course and the thing is you have the opportunity it is more effortful to not portray the truth it takes more effort to do that now it's less convenient but it takes more effort to lie than it does to tell the truth well um yeah that but that but that that Gap that difference between lying and telling the truth is is shrinking right because of the technology because of the access to to be able to do this so um you know and and but you're right I mean if you just if you pair it back to just s of the human nature right if you're you know but that's why if you're being interrogated you know you got to stick to the truth for the most part right you want to stay as close to the truth as possible obviously without veering off into something that's going to get your ass kicked but um the idea being because otherwise if you if you're lying you got to keep building on that you got to keep remembering and then you spend all your time thinking okay where was okay what did I say there I said that and then I said that and if you're dealing in an interrogation situation you know and that's what you're counting on you're counting on if that person's lying to me at some point they're going to get confused right and they're going to start mixing up that story and particularly if you you know in the old days you've got the ability for sleep deprivation or whatever it is right you can start messing with their mind a little bit then yes that's that's what you're counting on but so you're absolutely right in the sense that lying is much more difficult aside from the fact that you know usually if you make a mistake right the best we we were taught this at the agency right as as as being out in the field in operations I remember uh one of my first uh bosses told me when I landed on site before I went and and off to do whatever I was doing he said you know if anything happens if you make a mistake just come tell me we we'll sort it out right but just just tell me smartest thing right you can do to empower um young people with responsibility who are given responsibility to know that if they make a mistake come forward with it tell them and sure enough you know because of you know operations in the agency out in the field sometimes things happen so you know and he was good to his word right we had a bit of a goat rope and you know it it worked out but you got to be able to do that you got to be able to show it's not enough to say Hey you make a mistake come tell me we'll fix it you know if you kick their ass when they come Adit a mistake then you've you dis incentivize yeah exactly so you got to you got to follow through but you're so you're right sense yeah telling it's always the truth telling the truth is always the easiest path right but I guess my point being is in the world of disinformation now if your job is is propaganda or misinformation or disinformation it's becoming a lot easier right and a lot quicker a lot more efficient right to create that then in the old days if you were going to do a propaganda campaign right change hearts and Minds in some country right that you're you're trying to sway the people you I'm not talking about overthrowing the government you know those good old days are gone so I'm sorry that's not true we never did that sort of thing come on um but the point being is that you had that was a lot of work right because what were you doing you were using old media technology right you were finding journalists who might be inclined for whatever reason to write articles in a certain tone right and you create a network of them and you'd kind of pepper the the country with you know this this narrative and you'd build on that and hope that you having in the impact [ __ ] much easier now are you kidding me think about you you remember Cambridge analytica remember that so you know targeted um memes and and ads that were truned into cohorts and we have the uh maybe Hillary is a non- US citizen maybe Hillary is a you know all different ways that this gets split up you had all of these different uh um vulnerabilities and vectors of attack but there was still a person making a meme or making an advert or whatever so there was still human labor getting in the way of the rapidity of this whereas now you have both the ability for AI to deliver it in a targeted Manner and also to create it so you you can have Mike Baker profile on Facebook or Instagram or Tik Tok or whatever and it's who wouldn't want that no one wants that at all but and it's just you that it's targeted at so yeah I think uh yet increasingly we're going to see real information original sources you know maybe there'll be a way where uh being an actual person on social media will be a a rarity in the next five years maybe humans are going to be it's 99% of content will be created by Bots and every so often you'll see a tweet from you or or someone else that's real yeah and it's look if if you're if you're talking about a a a state right if you're talking about um creating something that's going to going to get past the scrutiny of of um you know Chinese Intel or Russian Intel or something yeah then you're you're working your ass off right you're creating a very hard back stopped elaborate elaborate cover um and it it's it's a lot of work but for the average person out there if what you're trying to do is influence a campaign or you're trying to do you know creating 10,000 uh fictional characters you know saying interesting things out there who then you know it's it's like the old concentric circles who then get 5,000 followers each and then it's I I guess but again it all comes down to the same problem right we can talk about all the you know the things you can do but it's individuals right it comes down to each person saying to themselves is what I'm reading or hearing or watching true you know and and I don't know how you you impact that right because human nature being what it is we're just you know we're we're consuming [ __ ] and then moving on path of least resistance yeah path of leas resistance and and to be fair most people are just trying to put food on the table or take care of their kids or whatever they're doing and so you can't ask them to spend a lot of time scrutinizing I don't have to time to be my own personal Snopes every single thing ex seeo it's not going to happen I again so I think that there's you know there there's a there's always going to be this this uh this this problem and not to say that it hasn't always existed right but I I do think the acceleration is pretty impressive right the speed of which you can create this this this fake right this deep fake or or this Persona that pushes information out there whatever it may be and uh the speed with which people are willing to consume it and pass it along right so that's something we're going to be battling I think for a while and and it will it impact this upcoming election of course it will and and future elections right why wouldn't you if I'm if I'm Russian intelligence or Chinese intelligence or the Iranians or who anybody who's got the resources why wouldn't I try to [ __ ] with the US yeah right so or or the UK or really it's not it's again it's not just a us-centric problem in other news this episode is brought to you by surf shark browsing the internet without a VPN is like going for a walk in a muddy field without any shoes on you can do it but it's ill advised and you might catch something you can protect your browsing online and get access to the entire world's Netflix library for less than the price of a cup of coffee per month if you're someone that works off public Wi-Fi networks like cafes or libraries or airports the internet administrator can see all of the traffic going back and forth between your devices and the internet and that is Tre bad you do not want this also your internet service provider is able to see all of your traffic and then sell it to companies that can Target you with ads websites are split testing you for prices on products that you're already buying and even flight websites are doing this too so to protect yourself from all of this plus to access the entire world's Netflix Library you can do that right now with one membership and it's across unlimited devices you can have it across your laptop your iPad your phone and even your smart TV you can get an 83% discount plus 3 months free by going to the link in the description below or heading to Surf shark. deod wisdom that's surfshark dodeals slod wisdom thank you very much for tuning in if you enjoyed that clip with Mike you will love the fulllength 2 and 1 half hour long podcast which is available right here go on there it is