For many years, I’ve lamented the state of Internet security. One aspect of this is that I’ve frequently wondered why people feel compelled to enter accurate real-world data about themselves into the cyber-world (i.e. the Internet), thus making the data found there more and more valuable to cybercriminals and oligarchs.
Now, we are facing a looming threat from AI at a much larger scale, but based on the same information that we freely share on the Internet. The only reason AI is looming in all of our lives is that we have collectively and graciously provided it (sometimes accidentally) with a huge corpus of relatively good data to work with (often at very little or no cost). There was a time when the rallying cry on the Internet was: Information wants to be free. The current day version reads more like Capital wants information to be free.
I would never suggest that people should provide false data about themselves to the authorities (i.e. organizations that you are legally bound to tell the truth to) such as the tax department, the courts, etc. But for everyone one else, especially the oligarchs and the cybercriminals, why on earth would you feel compelled to give them real, useful data to work with?
I don’t want to come across as being violently anti-AI here. AI has been here for some time, but in much smaller and more specialized areas (e.g. reading X-rays and spotting breast cancer better than humans) where it has frequently proven to be invaluable. What I am violently against is the intent of the current push to use general purpose AI to replace the livelihoods of huge numbers of humans and to put the resultant savings into the pockets of a few oligarchs. I can only imagine that this will just accelerate the arrival of the point at which the pitchforks and torches come out.
I have two humble suggestions:
-
At a personal level, take steps to poison the well with regard to your personal information, unless you are legally bound to provide accurate real-world data (i.e. renewing your driver’s license or filing your tax return). Use a bogus date of birth. Use an old or fictional physical address. Interchange your first and middle names. Just imagine if a cybercriminal searched the Internet for your information. If they came up with one name, one date of birth, one address; bingo, they have you! If they came up with a dozen values for each, how could they tell which were real?
-
At a government/regulatory level, we should pass laws and regulations that state that anyone operating AI should be able to demonstrate at any time that they had legal access to the data used to train said AI.