Thursday 28 April 2022

Agents In Society

If you had a super-intelligent General Artificial Intelligence and you asked it to produce as many paperclips as possible, the results could be catastrophic. It could destroy the Earth in search of the goal of optimising the manufacture of paperclips. This is an often used analogy designed to illustrate the dangers of poorly defined goals in artificial intelligence (AI) and machine learning (ML). However this paradigm of analysis has wider utility in consideration of society.

In a sense, though not directly what we would call AI, we have long been living with complex super-intelligences, and they have had direct societal impacts. For example the success, for a time, of both the tobacco and fossil fuel industries in mounting campaigns to delay or prevent actions on harms they cause is a result of systems rather than individual actions. The motives in each case are the driving goal of profit maximisation, and these goals have formed processes in the organisational systems that have led to the same behaviour, to obfuscate and to deny the evidence that threatens profit. These are examples of systems operating with a goal to the detriment of wider factors in society (human illness and global warming respectively).

A common, but poorly considered, view of such systems is that profit maximisation itself is the problem, such a view is often linked to belief systems antithetical to Capitalism. However these Anti-Capitalists have their own problems of unintended consequences being a result of poorly considered goals. A more considered view of profit maximisation is that this goal does not operate in a vacuum, in The West it operates in a framework of law drawn up by the legislature of democratically elected representatives. The Yang of legal limitations then interacts with the Yin of profit maximisation so as to avoid the goal or profit maximisation getting out of hand. Or ideally it should do.

The goal of the systems instantiated by Communism was to ensure equal access to the articles of consumption via state control of the means of production. This goal always requires totalitarian governmental apparatus (as does the Woke desire for Equity) and the outcome of seeking such goals has been writ large in the history of the 20th Century, with something of the order of 50 to 100 million deaths being attributed to direct and indirect effects of seeking the goal of equality. [Note 1 below.]

The above examples are hydrid human 'agents' seeking a goal whose underlying substrate are human minds and human bodies, albeit often enabled by machines, the hybrid aspect being the combination of actions of both individual and collective level actions in pursuit of a goal. 

We begin to get closer to similar actions of machine agents with the recent detrimental impacts of social media on humans. The agents here are systems that are a mix of humans and machines, seeking the goal of increasing engagement in social media platforms, which has led to a general increase of psychological problems such as narcissism and the fragmentation of society. In tandem with the spread of social media there has been a severe increase in depression, particularly amongst girls (fig 2 of that source).

Considering the general vulnerability of society to falling prey to societal manipulation by agents seeking goals, as evidenced by the above repeated episodes. And considering the lack of a reliable defensive immune response within Western Liberal Democracy, similar to the body's biological immune system. This evidence does not bode well for societal resilience to handling highly intelligent machine agents in our midst with goals that generate behaviour that prove problematic for society.

However the starting point for a defensive response against rogue agents pursuing goals with dangerous consequences always whittles down to one key factor: Freedom to speak and to criticise. This creates a feedback of information that allows goals and their seeking mechanisms to be critically analysed. Only such criticism and revision of goals can lead to corrections.

If we lose the ability to speak out, because perhaps we get 'cancelled' if we dare to do so, then these societal agents continue to work towards damaging goals and deleterious impacts may occur. This is why the recent acquisition of Twitter by Elon Musk is a positive event. Twitter is a major social force for discourse and it has been firmly under control of the Woke Totalitarian movement. To see that control loosened is a hopeful prospect.

[Note 1.] This is not to say that the goal of individual freedom is not without its problems. The ideal path probably takes something from collectivism and something from individualism, however to ask what is that path (singular) implies Utopian thinking, and Utopian ideas are always toxic and dangerous. The truth is that the realistic path is one of continual review and revision. The world is dynamic, why expect simple answers that apply at all times in all circumstances?

No comments: