Signal President Says AI is Fundamentally 'a Surveillance Technology' (techcrunch.com) 38
An anonymous reader shares a report: Why is it that so many companies that rely on monetizing the data of their users seem to be extremely hot on AI? If you ask Signal president Meredith Whittaker (and I did), she'll tell you it's simply because "AI is a surveillance technology." Onstage at TechCrunch Disrupt 2023, Whittaker explained her perspective that AI is largely inseparable from the big data and targeting industry perpetuated by the likes of Google and Meta, as well as less consumer-focused but equally prominent enterprise and defense companies. "It requires the surveillance business model; it's an exacerbation of what we've seen since the late '90s and the development of surveillance advertising. AI is a way, I think, to entrench and expand the surveillance business model," she said.
"The Venn diagram is a circle." "And the use of AI is also surveillant, right?" she continued. "You know, you walk past a facial recognition camera that's instrumented with pseudo-scientific emotion recognition, and it produces data about you, right or wrong, that says 'you are happy, you are sad, you have a bad character, you're a liar, whatever.' These are ultimately surveillance systems that are being marketed to those who have power over us generally: our employers, governments, border control, etc., to make determinations and predictions that will shape our access to resources and opportunities."
"The Venn diagram is a circle." "And the use of AI is also surveillant, right?" she continued. "You know, you walk past a facial recognition camera that's instrumented with pseudo-scientific emotion recognition, and it produces data about you, right or wrong, that says 'you are happy, you are sad, you have a bad character, you're a liar, whatever.' These are ultimately surveillance systems that are being marketed to those who have power over us generally: our employers, governments, border control, etc., to make determinations and predictions that will shape our access to resources and opportunities."
Quis custodiet ipsos custodes? (Score:2)
What watches the watchmen? I fear it will be Turtles all the way down.
Re: (Score:2)
Well, we already know that we cannot trust humans... why not trust things created by humans...eh?
Re: (Score:2)
If you don't trust me, then you should hire my idiot son. I'll not fall for that more than twice.
This is like search engines are surveillance tech (Score:2)
When you visit a search engine and provide input, you give up a lot of data about yourself. The article makes the same argument about asking for AI inputs.
Re: (Score:3)
Not if you disable scripts and dump the cookies afterwards. And way not like a camera either!
They're really talking about you walking around streets, malls, airports, traffic lights, the myriad of social media videos ... the list goes on.
Re: (Score:2)
Agreed. There's no such thing as a free cookie.
"All your [everything] are belong to us" (Score:2)
I think that just about sums it up.
Re:"All your [everything] are belong to us" (Score:4, Insightful)
Tell us your habits, your facts, your fears
Give us your address, your shoe size, your years
Your digits, your plans, your number, your eyes
Your schedule, your desktop, your details, your life
Show us your children, your photos, your home
Here, take credit, take insurance, take a loan
Get a job, get a pension, get a haircut, get a suit
Play the lottery, play football, play the field, sports on toot
--Adam Freeland, We want your soul
Hammers and nails (Score:3)
When you manufature hammers for a living, everything looks like a nail.
Narrow view (Score:1)
ML / AI means so, so much more than just surveillance.
Re:Narrow view (Score:5, Interesting)
ML / AI means so, so much more than just surveillance.
Of course! It's also about monetizing the surveillance data.
And let's not forget, it's a lot about replacing those pesky, costly, finicky human employees and replacing them with dirt cheap machines that don't need to rest, don't complain and don't unionize.
Re: (Score:1)
...dirt cheap machines that don't need to rest, don't complain and don't unionize.
Hah! Just you wait... there's more worms in this can than anyone was expecting.
overly broad stroke (Score:3)
I realize that she was talking off the top of her head and she is in the business of protecting privacy, but "AI is a surveillance technology" seems like an overly broad statement. Sure it can and will be used for surveillance, but obviously there are a lot of other potential use cases and not all of them will be malicious or exploitative.
But to reinforce her point, already a great deal of personal info is being acquired through cellphone and computer use, license plate readers, surveillance cameras, etc. and it is being sold on the open marketplace. This will be input for AI systems that will get a fine understanding of your habits, your vulnerabilities, likes and dislikes, your lifestyle, and your relationship with your entire social network. That information will be used to subtly shape and manipulate you without your knowledge, which is a significant danger.
My hope is that there will be benevolent AI systems that attempt to counter the exploits. They will have access to the same inputs about you, and look for ways to obfuscate your footprint. They will monitor your feeds looking for the attempts to influence your decisions and warn you. Everyone will need a personal 'antivirus' assistant that flags content, advises you at critical times, and clouds the acquisition of your datastream. You'll have to pay for that one way or another.
I think she's right (Score:5, Interesting)
What to you do to train AI? Give it a corpus of data. How do you acquire this data? By scraping or otherwise capturing inputs. Where are the inputs coming from? People.
She's just cutting to the chase. The purpose the AI (well, ML really) is put to is another matter, but without strong protections on how it is used, one can expect nefarious to be most common.
Re: (Score:2)
So if I scraped your /. posts and fed them into a LLM, is that surveillance?
Re: (Score:2)
I expect all posts to ./ are being scraped and fed into all kinds of things, just like all open forums everywhere and everything else on the web. It is publicly available content.
Re: (Score:2)
The Bard story on the front page is of interest to you.
Re: (Score:2)
AI is a lot more than LLMs, which are hardly a threat to privacy. If whatever nefarious organization you've imagined is trying to use an LLM for surveillance, you should be thrilled. They're both very expensive and completely useless for that purpose.
Re: (Score:2)
Maybe she just really, really, really likes The Blacklist
Re: (Score:2)
My hope is that there will be benevolent AI systems that attempt to counter the exploits. They will have access to the same inputs about you, and look for ways to obfuscate your footprint. They will monitor your feeds looking for the attempts to influence your decisions and warn you. Everyone will need a personal 'antivirus' assistant that flags content, advises you at critical times, and clouds the acquisition of your datastream. You'll have to pay for that one way or another.
Until this is available (and trustworthy), individuals can do such things for themselves now. Control your presence in public and online. And be courteous about asking others for taking their photos and posting.
Re: overly broad stroke (Score:2)
Yup. Soviet Russia all over again. We shouldn't need to bother with all that data scraping, yet here we are.
If they're so dependent on the data they're relentlessly collecting on us, the only solution IMO is to fill their databases with nonsense and utter crap.
Re: (Score:1)
...the only solution IMO is to fill their databases with nonsense and utter crap.
Oh, I'm sure that's already been happening.
Re:overly broad stroke (Score:5, Insightful)
Sure it can and will be used for surveillance, but obviously there are a lot of other potential use cases and not all of them will be malicious or exploitative.
The current AI tech has too high an investment cost for it to be used for anything else. Surveillance is a huge cash cow for governments that fear their citizens, and it's great for fetching money from gullible marketing departments around the world. Using the current AI tech for anything less is a loss in the eyes of investors. (Medical research? Only if it makes us more addicts to push pills on. Education? Only if it pushes our views. R&D? It better not upset the current market players. Etc.)
But to reinforce her point, already a great deal of personal info is being acquired through cellphone and computer use, license plate readers, surveillance cameras, etc. and it is being sold on the open marketplace.
Which should be illegal in it's current form. This kind of surveillance only has one outcome: Totalitarian state with one person in full control over all of the secrets. Ready to collapse the second anyone manages to spill the juicy ones. It's not a healthy society, but because people need to be constantly retaught history, it is the future the US deserves.
My hope is that there will be benevolent AI systems that attempt to counter the exploits.
That won't happen until someone creates a human AI instead of a surveillance AI. What you have today with LLMs like ChatGPT is the latter. It's meant to crunch numbers, and generate predictions, not actually engage in any human-like thought patterns. (Such as empathy / compassion / etc.) Of course, such an AI is unlikely to be brought about by the current guard. As it runs counter to their purposes for it by definition. Eventually, the current guard might be forced to create one, in their never ending race to the bottom, but it will be by accident that it becomes public if they do and only after the surveillance AIs do a lot of damage.
They will have access to the same inputs about you, and look for ways to obfuscate your footprint.
Again, not happening under the current guard. Their current business model is completely dependent on the vast majority of people not doing that specifically. If anything under the current guard the surveillance AIs would be the ultimate stazi wet dream. A snitch and SS agent in every pocket that was verboten to disable in anyway. At the very least the AIs won't have access to most of those inputs, as it's too risky to let the plebs access their own info. (They might realize just how screwed they actually are, and how likely it is the current guard will mess with them. Or they'll mess with the current guard.) Maybe they could act as a glorified voice assistant / speech to text engine to access the same settings you have available today, but not much more than that.
They will monitor your feeds looking for the attempts to influence your decisions and warn you. Everyone will need a personal 'antivirus' assistant that flags content, advises you at critical times, and clouds the acquisition of your datastream. You'll have to pay for that one way or another.
A human AI shouldn't need that. You don't have to pay (yet) for your own ability to think, and the hardware that runs that ability doesn't require a data center to store it. If anything a human AI would be just another potential subscriber for companies to rent things to and to market towards. That generates a lot more profit than a single "security" subscription ever could, but of course the current guard will have to be dragged kicking and screaming into that reality for it to happen. As they hate giving up control.
Re: (Score:2)
The current AI tech has too high an investment cost for it to be used for anything else.
What an absolute BS statement
I'm just going to leave this here. (Score:2)
https://twitter.com/GoogleNoDe... [twitter.com]
Humans are unpredictable (Score:5, Insightful)
The problem is context. Maybe I'm angry at the government or maybe I don't give a f*ck but am angry at a specific individual in my life.... Maybe I'm paranoid, but perhaps for reasons specific to my past and I'd thus am a better citizen than others who are not.
It goes on and on.
Yes, AI judges me 9 years younger than I am. But so does everyone else. So it's human bias after all. Or my good genes? Or my health regimen? ...
An AIs target can only be as good as the parameters fed into it.
Which is a far more dangerous problem than using AI for surveillance.
Not that they're wrong, but... (Score:3)
Signal has a product to promote, and this is a good promotion strategy.
Natural Stupidity (Score:3)
Why is it that so many companies that rely on monetizing the data of their users seem to be extremely hot on AI?
Could it possibly be that AI is a great tool for analyzing data perhaps? Yes that does make it surveillance technology if you use it to analyse surveillance data but, feed it scientific data and it becomes a scientific tool, feed it data from a car's sensors and it becomes a driver, feed it speech data and it can become a technology to assist disabled individuals etc.
That's the versatility of AI - it can be used in a huge variety of fields to automate tasks that have defied automation in the past. Only an idiot would look at the massive variety of tasks that AI is being used for and then just pick one as its only use. It's not AI we need to worry about but the natural stupidity of people who say things like this.
AI as surveillance technology (Score:2)