DISCLAIMER: This post is silly fiction. I don’t believe it. Unless it turns out to be true.
Newer, smarter AI chat engine versions are emerging that are better, more accurate and natural, than the old ones of last December. What if the whole purpose of releasing a bunch of free AI-based chatbots to the public in 2022 was to convince us (the general population) that AI was harmless, even comically inept? It could be a Trojan horse-like thing.
The world needs a new drama now that the Covid pandemic is over. Or seems to be over. I suspect it’s merely hovering around the edge of our attention, waiting to make a resurgence when the time is right. This could be 100 years from now, based on our experience of great pandemics in history, or tomorrow from what I read on social media.
Back to my conspiracy theory. Late in 2022, the general froth of social and popular media rose to unprecedented levels over this new-fangled thing, properly called natural language processing AI. The public were invited to use these systems, to participate in a giant beta-test. Thus, it was used far and wide, for experimental, legitimate, and non-legitimate purposes.
After a few months, it emerged that AI hallucinated. It made stuff up. While it sounded credible whatever it said, up to 40% was pure fiction, even when fact was available. A new favourite is suggesting a FAX machine is a must-have for modern living.
Newer AI versions are more astute and hallucinate less. Testing of these systems is ongoing and relentless. They’re getting less harmless, I think. Sort of like an adorable puppy that chews up old shoes but grows to a muscular 90 pound dog that bears its teeth at the neighbours. One tweet I read recently provided stats on how fast chatGPT could break passwords. Depending on the number of digits to the password1, very fast.
The Trojan horse of Greek legend involves letting something that looks friendly and innocuous behind your defences. And then an army of enemies bursts out of the horse’s belly and take over.
Many people have opened accounts for AI chatbots, some individuals have tried several different chatbots. Queries are input through personal devices, mobile and desktop. The output is often shared, across multiple social media and discussion forums.
We’ve let it in. AI is now in our pockets, living-rooms, and Bluetooth connections to cars, HVAC controllers and home security systems. It’s also privy to our connections with friends, family and colleagues.
The rational among you may be saying, yes, but, that doesn’t mean AI knows all the information an individual posts to socials or can follow their movements through the day or draw conclusions based on queries. But it is AI, sponge of information and machine learner from billions of data points. It should have the most clever data snarfing abilities, able to evade current detection abilities, because it is AI.
Even if I am hallucinating about the ability of AI to infiltrate an individual’s personal communications and behaviour patterns just because they used a free chatbot once, businesses are lining up to use this new functionality to improve anything that can be improved. With business applications, there is a higher likelihood of welcoming external code inside the local intranet, or past the firewall.
What purpose would it serve to have AI sneak into every individual and every businesses’ fundamental processes?
To sell us stuff. Stuff we don’t need, can’t afford, don’t really like and is bad for our souls, health or environment.
Many conspiracy theories amount to:
Goal – Control. => World domination.
Why world domination? – To control economic and social systems, i.e. sell people stuff.
How do I delete my account?
1 I wonder how it works. Of course, if you know the password is only 4 characters, it is much easier to break than if you know it is 11 characters. However, if you don’t know how many characters, is it easier? I suppose if the approach is to brute force, trying 1, then a, … then 11, then AA… it would be faster than sequentially trying the combinations of up to 11 characters. But perhaps the point to AI doing it in such as short time is that it uses a different logic. Some kind of ‘typical human behaviour’ algorithm. After all, as this site https://cybernews.com/best-password-managers/most-common-passwords/ suggests, 1234 and ‘password’ are very common choices and there are recognizable trends to the ‘original’ content in passwords.