Quite simply because it's fine when the Government and Police can be trusted, but recent actions by both have proven that for certain demographics that that is far from the case.
But not trusted to do what ?
AI isnt Judge Dredd or Robocop, its not going to come through the roof of kings cross and grab someone, though to the criminal it may feel like that if they find themselves surrounded an hour or after committing a crime in York.
its going to look more like this..
They’re built to record, not intervene.
www.theverge.com
even thats flawed, one ended up in an indoor pond before now.
its an information tool. A human is going to use it like any other tool.
If theres a threat here, its to Googles search engine, as it will likely replace it. Thats why they are spending millions on ensuring they have an AI engine.
If AI finds a candidate, the legal evidential process is still needed to convict (Ie dna, witnesses etc).
The difference should be faster and more accurate search, retrieval & relevence. It will however rely on much greater levels of data integration than we have today and use “thought”* to model, where as Google today can only share what it finds.
it has all kinds of potential, in medical, farming, commercial, financial etc.
A lot of fear is about military, but even the worst dictator is going to want their own control over the system, to ensure it doesn't turn against them.
AI itself has a weakness for power.. not control, but energy. Its weakness.. if AI gets ahead of itself, stop the flow of electrons by flipping the “off” switch on the wall.
* thought today is structured code.. if x=1 then A else B etc.. AI will be able to adjust, ie “Whatif“ and use a wider range of variables, re-evaluate and share its learnings.
For example..
Remove all ticket barriers, all ticket machines.
People facial register, and link credit card details.
Flow of travel and billing is automatic, including any adjustments.
TTIs now get advance notification to focus on mis-matches, unregistered and notified of known offenders.
Trainseats could be allocated automatically, re-route/notifications for late cancelled could be calculated automatically.
Further, AI could reccomend which stations to skip/short terminate based on loadings, rather than a controller making the same decision. It may also be able to reccomend changes to timetabling, load factors etc, based not just on revenue.
it would mean the end of TFLs Medium sized business that is penalty fares collection, but isnt that the morally right thing it should be… ? to me tfl seeking to preserve that PF revenue business model is a bigger threat.