From Alex Stamos, director of Stanford University’s Internet Observatory, on the capabilities of modern AI language tools:
I think we’re entering an incredible era of bullshit.
I can hardly wait.
27 thoughts on “Quote of the day: A golden age of bullshit”
Comments are closed.
See your previous post: "Republicans, it's conspiracies all the way down"
No AI needed.
I was told it was to be a dangerous possibly deadly era...oh wait. We drown in the bullshit, right?
What's coming is worse. We had a preview of it with the hacked broadcasts in Russia with a believable looking Putin announcing on camera of an attack, mobilization, and martial law in Russia.
There will be widespread videos of officials, including secretary of states, giving out fake voting information or doing illegal activities. Some will be released to sow chaos in the middle of voting, sending people to wrong places or with incorrect instructions that will invalidate votes.
Officials will be caught off-guard, incapable of responding in time to prevent its spread. A lot of it will be spread organically and through bot accounts on social media, especially birdsite.
We entered that in March 2020 with extraordinarily stupid lockdowns that then led to 18 month school closing, mask mandates and then vaccine mandates for an experimental drug that offered no benefit to anyone healthy under 40. Thanks, Dems.
Speaking of disinfo.
...showing once again that Artificial Intelligence will never surpass Natural Ignorance.
“Entering”?
AI will not make me believe anything, but I do worry about people who are immersed in social media. They are already immune to common sense. It would be fun to fuck with them. And they really deserve it.
"AI could supercharge this labor-intensive process by automating the content creation and mass-testing of what content is most engaging, which could be measured by viewership, e-mail responses, and donations. AI could also use the most engaging content to trigger the social media architecture that features the trending posts more prominently, giving false information undue and unearned attention."
Since I don't use social media and have ad blockers keeping all the bullshit at bay, I'm safe for now.
That is a little like saying that because you keep your gun in a safe place you are safe from being shot.
The problem is not what the people here will do. The problem is what the Qanon believeres will do once they see fake but realistic videos confirming their conspiracies.
They're already seeing fake and unrealistic videos confirming their conspiracies and believing them.
If they really believe them you would already have a civil war. At the moment the majority just pretend to believe.
With realistic fakes, that may change.
The problem is you aren't as smart as you think you are. You are kind of like the guy driving 80 mph, passing people on the right, who thinks everyone else on the road is a bad driver.
It's not about making you believe something, it's about the ability to recognize patterns that you don't know you have and using them. It's something we already do, and fall for, so AI could make it worse.
It's not necessarily QANON crap, it could also be the D's finding ways to manipulate people into more fundraising or other ethical gray areas.
Or, maybe it won't make any difference. Things like this have weird blowbacks, so I guess we'll have to just wait and see.
"AI will not make me believe anything"
That's like saying "ads don't affect me." It *might* be true, but what are the odds?
Heh. Depends on what you mean by 'affect', doesn't it? If I see an ad for $2 off a bag of my regular coffee, then yes, it will 'affect' me. If I see an ad for impossible coffee, no, it won't. So which sense of the word do you mean in this context?
I have, by the grace of God, the gift of barely registering ads. My wife and I have sat in front of the TV and she would say something about the commercial we just sat through, and be astonished when I had no clue what was in that commercial.
I have had marketing types assure me that I am in fact subliminally influenced by those ads, which I am assured are very effective. But of course marketing types would say that.
Same here. Based on the three of us in our house, I'm the outlier. But when an ad comes on, I instantly tune out. Why waste a precious few minutes of my attention on somebody trying to sell me something I don't want anyway.
All of the largest corporations on Earth spend billions and billions of dollars annually to put their ads where we see them. Why do they do that? Is it just to let you know when your regular coffee is on sale? Permit me to suggest: no. Ads are psychic depth bombs that push our behavior in directions that favor them. When you say, "I'm immune", the ad guys high-five each other.
Remember the rule for streaming TV: "free with ads" is an oxymoron. You WILL pay them for what you watch, either directly or indirectly. Messrs. hersh and Harbin (above) should consider that an ad which is quickly forgotten can still leave a mark in the subconscious. Either every corporation on Earth is stupid for thinking ads work, or we're all stupid for thinking they don't. Which is it? Keep in mind that corporations have actually studied this question in depth, as they study all large investments.
It is posisble for ads to be useful for the advertisers even if they affect only part of the population, so it is quite possible that there is a fraction of thepopulation that is immune to ads but they are still viable.
Though it is also true that people that think they are immune to ads are not always actually immune.
Well that's rather convenient: we know advertising works because we spend money on it because it works...
There's no evidence that online advertising works.
Even if a publicly traded company knew its advertising program would not likely increase sales they would still do it because the perception might move the stock price which is what really matters. The beauty contest theory.
Heh. I will cheerfully concede that *microtargeted online ads* - the actual topic of your linked article - are ineffective, though I would argue that the problem there is the microtargeting.
Or maybe no corporation has ever thought to just tell their stockholders about the multiple billions that they could redirect away from advertising, towards dividends, R&D, and actual profitable enterprise.
That's now how the game works though. The goal is to increase the value of the stock. Real innovation that's scalable is very rare and costly. You can't count on that.
It's simply less important that a McDonalds Superbowl ad convince you or I to have a burger (like neither of us know about this restaurant?) than it is that the ad makes the guys on the earnings call from Morgan Stanley feel good about the brand and their bank making huge investments in McDonalds. That perception is what actually matters.
Since I don't use social media and have ad blockers keeping all the bullshit at bay, I'm safe for now.
If I never see the ads, how can they affect me?
Your comment reminded me of this:
"And Darkness and Decay and the Red Death held illimitable dominion over all."
There is a second layer of BS awaiting us: BS generated by language models. They have zero BS detector built in.
I participate oseveraccasionally in forum for violin players. There somebody posted about CatGPT: It had mentioned biographical facts about one the great violinists of the 20th century that had been unknown to the person who was very impressed. In response several people pointed out that ChatGPT was not a reliable source of facts.
One person published the answer of ChatGPT to a question about "thumb vibrato" (for non-violinists: There is arm vibrato, finger vibrato, hand vibrato but no thumb vibrato. It is physically impossible). The bot came up with a long paragraph about thumb vibrato, its history, which violinists have used it etc. etc. It was hilarious.
In other contexts this sort of BS could be very dangerous: It could be snake oil medical advice, it could be political BS, it could be dangerous tips as to how to fix something about the electric wiring etc. And there is not even a need for a bad actor.
oseveraccasionally
CatGPT
When LLMs get around to vacuuming up this blog and its comments to train on, look for these words to become part of the language. (CatGPT sounds like it could become a Friday afternoon feature right here.)
I think it's worth thinking about different categories of "bullshit."
1. BS produced by malevolent actors (fraud and disinformation)
2. Errors made by repeating bad source information
3. "Hallucinations," as they're calling it, when LLMs just make shit up
The industry argues that the cause of #3 is a mystery, and according to some, unknowable bc of the nature of how LLMs work. I think they're either lazy or hiding something, and the explanation they give is ... bullshit.
It doesn’t matter if you believe it or not. Think of Labyrinths by Jorge Luis Borges. The library is full of books that both prove and disprove everything. There is no way to know what is true.
Pro-DeSantis group unleashes video ad with fakes created with AI.