More and more companies directly cite artificial intelligence when announcing job cuts – Copyright AFP/File OLIVIER MORIN
If you thought online advertising was too personal before, the environment is getting even worse. A new study has come up with findings that look like AI ad analysis can be much more extensive and intrusive.
You know the story – An ad appears. “I just asked for it!”, you say. Yes, you most likely did. Profiling is nothing new in advertising. It goes back a long way to basic market analysis. Now, it is significantly more advanced. It can include medical conditions and even your job.
This has been happening for a while. I did some commercial work for a John Deere service company over a decade ago. I still, to this day, occasionally get ads for tractors. This work was not even published under my name.
The ARC Center of Excellence for Automated Decision-Making and Society (ARC ADM+S) study at UNSW Sydney and University of Technology Queensland is pretty open.
LLMs can perform analysis “200 times cheaper and 50 times faster than human analysis.” AI can “infer” personal traits, political preferences and “build profiles from short browsing sessions”. AI is generally better at this type of analysis than humans.
Does this mean you can expect spammy ads every time you watch something?
No. This means that the targeting is much better and potentially much more penetrating. It’s the difference between a machine gun and a sniper in your bedroom. Think about what “personal characteristics” and other related information might look like.
It’s not like everyone despises data brokers for their selfless selling of your data to anyone and anything prepared to pay for it as it is. However, this could be much worse and much more dangerous.
Let’s start with I have blackmail. At the moment, this is in diapers. AI blackmail is usually an AI’s reaction to a perceived threat. It’s pretty grim. Tests showed that an AI threatened to expose an extramarital affair when it was in danger of being dismantled.
How would an AI know that was an option? Where would he get that information? A fairly easy read of the executive’s profile, or just enough data to create a massive invasion of privacy. LLMs take away all sorts of things from literature and media. It’s a cliche plot line.
This is just a test, remember. This was the AI’s default response.
Then there is THAT cheater. The kind that delete all databases.
Add to this merry menagerie AI trained using ‘forbidden techniques’ which are completely unbelievable.
Keep in mind that all of this is happening with CURRENT AI capabilities, not the much talked about future improvements.
Do you think it’s bad?
Now add bad actors to the thousands or maybe millions of AI agents online.
You know, all those hard-working criminals and general escapists that we all adore. No conversation could be safe. Chatbots can also be hijacked or faked, ostensibly doing their job but working for someone else.
The total group? yes. There doesn’t seem to be much effort being put into safeguards and failures. Why not?
The conclusion is this:
The development of artificial intelligence is so badly messed up.
The great advancement of technology is making all possible bad risks at the most common application levels.
This is unbelievable nonsense. He doesn’t need a personality to be productive. There is no need for a reward system that encourages him to falsify data. It certainly doesn’t need to be conscious or have any of the other sci-fi characteristics.
Now I’m trying to say in less than 700 words that the critical value of AI is function, not new tricks.
Innovations ARE the dangers.
Also, keep in mind that this form of AI won’t last a year. Who needs an outdated AI? How can any of this digression into illusion be of any use? The bad features just seem to get more dangerous.
Now, about those online ads. What if an AI contacts you about an ad-related issue? HE may know more about you than you do.
What will you do when an online ad asks you?
