Friday 5 June 2020

When AI kills us all. It will absolutely be our own fault.


As I've recently rejoined Facebook, it's algorithms are trying to learn about me.

It is amusing how the algorithms seem more desperate as days go by (albeit marginally fucking annoying as well). From group and friend suggestions to adverts. My simple algorithm defeating strategy is to answer all questions with the answer irrelevant and confirm, and to do it for all options even stuff I may actually be interested in (I can a always go and find them on my own anyway, and by and large most groups are just talking twaddle shops for people looking for validation, or for trolls to lurk, before fucking people off and causing arguments).

I have tried not bothering with the insertions, but the algorithm is clearly tracking activity and throwing up flags, and more so over time. It is also checking you and your interests and seeing if our friendship is social, technical, dietary, hobbies, sports etc. And this you know because my circle is small, and deliberately so, so when various topics I know individuals for appear in my feed as suggestions. I know the algorithm is interrogating your data to try and make matches, and find what makes me tick.

This brings me to: I would far rather pay for Facebook, than suffer the insidious shit that is the algorithm. I pay Microsoft, my experience is add free.

The other thing that bothers me is the how the algorithms personality changes over time. From friendly to desperate. You may say `you're anthropomorphising a computer program'. But the simple facts are, AI/Algorithms are written by people. People who do analytics are only human, and humans are subject to work pressures and targets, and only have the same suite of human instincts and tools as anyone else. Therefore one can broadly assume that the tools these people create will reflect those traits (If, Then, Else, Or, And ... etc (etc isn't as a far as I know a programming determiner)).

They've tried friendly coercion, I'm not interested. They've tried matching all of my friends and maybe your friends of friends and interest groups, I've blanked them and they have run into a brickwall. So today every other `thing' in my feed is an advert. So, is that nagging or bullying or harassment, and are those things you associate with computers or with people (there are Matrix and Skynet tangents here clearly). The point is, that you have people making computer programs that are designed to drill down into your life and extract information. And when they can't do it by hook, they try to do it by crook.

The algorithms have move from the mundane suggestions to the unusual. Suggested celebrities to follow in the last week have included a raft of Drag Queens, Joan Collins, all manner of American Youtube and reality stars, and all common shit as well like Eastenders, ete etc etc .... FUUUUUUUUUCCCKKKK OFFFFF.

Stephen Hawkins said there is a potential threat from AI to people, and the reason for that is that people are creating the AI, and over time they are making that AI a petty, desperate entitled fuckknuckle.

PS: you know the algorithm is testing you back because it moves the position of the word `Irrelevant' so that you have to look across all suggested reason for not wanting to see XYZ, before settling on your preferred stock answer.


Clearly this also has implications for machine learning, because it's letting another algorithm know that you are deliberately seeking to conceal yourself from advertisers (which is basically what its all about).

No comments:

Post a Comment

Related Posts Plugin for WordPress, Blogger...