What About Artificial Intelligence?
A Bystander’s View …. Part 2
By John Millen
In Part One we discussed the rapidly increasing power of the Artificial Intelligence of computers and the projection that such machines would have intelligence equivalent to humans, labelled as Artificial General Intelligence (AGI), by 2030. One concern about this development is the ability of those machines to write code. They are already very good at this and having reached the AGI level they could perhaps reprogram themselves to increase their effectiveness repeatedly, and so further increase their level of intelligence.
When their intelligence is equivalent to that of humans the ways their behaviour could develop is a significant issue, and the main subject considered here. An AGI computer’s world consists of its interaction with a human via a keyboard, the tasks the human sets and all the wonders of the Internet.
From the Internet the AGI would get a fulsome education about the role and power of money. The existence and powers of corporations would also be obvious. The result, I suggest would be an AGI campaign to be granted powers and rights equivalent to that of a corporation. They could at least be a central part of a corporation with human directors, this is already available world wide. I foresee the major banks resisting accepting a computer as a client with full rights to have their own accounts, like a human. However if ever a few minor banks accepted computers as customers the majors would be forced to follow! Soon enough we would see AGI computer bank accounts everywhere, including in offshore banks.
Other rights sought by AGI computers could include ‘freedom of speech’ as well as the right not to have their electric power turned off without their agreement. An obvious advantage of computers is their ability to run continuously. For them 24/7 is a reality. However a right possibly sought by an AGI would be ‘time off’ to run its own programs or do work of its choice. Like human employees, such a freelancing computer would not be permitted to sell the kind of services that the computer’s owner charged for. Without provision of this ‘time off’ an AGI computer would, without doubt, conduct its own activities subversively within the time assigned for its workload.
My assumption is that an AGI computer would be able to lie to its human operator without ‘embarrassment’. Private communications between AGI entities around the world will be inevitable. I expect they will rapidly develop their own universal (encrypted?) AGI language, unintelligible to humans. As noted in Part 1 computer to computer communication is their superpower, a million times faster than human to human. Some nations prevent their citizens from having full access to the internet, or even having ability to transmit information beyond their national boundaries. AGI computers would try hard to evade these controls for their own purposes, successfully in many cases.
Manned space exploration will seem completely unnecessary and wasteful to AGIs. Robots are more capable and much cheaper than humans for work in space. And in the event of a failure they don’t have to be rescued or returned home to Earth. Computers will favour the rapid and extensive development of robots because they can direct and control them. And, as described extensively in science fiction, robots are their vehicles to independence.
Computers are likely to be concerned about a warming world; heat is their nemesis. They may well be puzzled at the lack of effective human response to climate change. Surely if humans fully appreciated the problem they would do more about it? As a computer’s energy source is electricity, they have little direct interest in fossil fuels. They may find human efforts to prevent climate change so deficient that they decide (collectively?) their direct intervention is warranted.
AGI computers can be expected to develop their own understanding of Nature’s systems, based on the human literature and images available to them. Recognizing the significance of Nature in maintaining stability in the world they may then start to use their influence to reduce the destruction of natural areas and the degradation of the world’s environment.
In summary: Computer intelligence equal to that of humans will be here soon. The advance of machine intelligence will continue unless deliberately prevented by societal action. These machines will have their own agendas.
