The Appropriate Response to AI
In a TED talk in 2016, Sam Harris suggested that we are unable to marshall an appropriate emotional response to the growing power and utility of artificial intelligence and that, given our propensity to continually improve our intelligent machines, the emergence of machines that are vastly more intelligent than we are is inevitable. Fast forward to 2022 and fields of human endeavour that were thought to be impossible for machines to achieve are being overrun with increasing rapidity.
We persist in believing that human intelligence and consciousness is somehow elevated above simple atoms, molecules, neurons and electrical impulses. It’s not.
Writers like Daniel Dennet have long argued that consciousness is fundamentally algorithmic and has been achieved in humans through eons of evolution from the simplest of building blocks. Other writers have proposed consiousness in animals, even simple systems — consciousness is a matter of degree. Furthermore, consciousness does not need to achieve anything like humans levels to threaten human activities that were traditionally thought to rely on an intrinsically human “spirit” or “soul”.
For example, MidJourney produces startling imagery from simple text prompts. ChatGPT can write country music lyrics about your dog dying from eating cooked chicken bones, Sharespearean sonnets about peanut butter and rap music and an interesting discourse on whether Jesus was a real person and had any decendents (some of the things I’ve conversed with it about :-). I know chatGPT is not conscious in the same way I am (yet). It’s just a sophisticated algorithm trained on a set of historical content, but then, intellectually, so am I.
In this head-to-head comparison of art generated by a human graphic designer and the DALL-E image-generation AI, the results seems to be a win for the human designer overall but in the process caused a lot of debate among the participants and, in one case, DALL-E seemed to meet the brief better than the human designer. It should be pointed out that DALL-E took a matter of seconds to generate this output compare to hours for the human designer. If the DALL-E output is considered good enough it could already be making some graphic designers redundant without any discussion of its level of consciousness.
We persist in believing we are fundamentally “special” to our detriment.
If you find yourself making a statement to the tune of “an AI can’t do this thing that humans can do”, just add the word “yet”. Sam Harris is emphatic that we have no idea how long it will take us to be able to create the conditions necessary to produce superintelligence safely. I’d suggest that, in early 2023 and with the release of chatGPT, we’ve crossed a threshhold with exponential speed and have no idea how much time we have left to deal with it. Imagine the fabled race between the tortoise and the hare: the AI hare has hit the start line at speed and continues to accelerate while the human tortoise is left blinking in the dust. Perhaps this tortoise should give up this particular race and find one more suited to it?
But humans are still in control of AI, right? It only does what humans tell it to do, right? It depends which humans you’re talking about. Our global capitalist system is driven by a small class of capital owners for profit. Businesses are constantly looking to remove costs from their systems to increase profits. The largest cost for most businesses is employees, so if employee costs can be reduced or eliminated, increasingly through automation and AI, profits go up. The humans currently in control of AI are those who will profit from it, and everyone else is left fighting over the remaining jobs it leaves behind.
The common argument against automation killing jobs is that new jobs are inevitably created to manage the automation (also, managers tend to like having people to manage to justify their roles, so new bullshit jobs will tend to get invented to fill the gaps). However, the profit motive will ensure that, over time, costs will come down and the remaining jobs will be those that an AI cannot (yet) perform.
If previously complex operations, requiring highly skilled workers, can be simplified so that lower-skilled workers, preferably in low-wage economies, can be used to do the work, this will also lower costs.
The fundamental challenge for capitalism in the age of AI is that businesses need to be able to sell goods and services they produce to make a profit, so something or someone needs to be able to buy them. If the class of skilled, highly-paid workers as consumers is being hollowed out and replaced with lower-skilled or lower-paid workers, who will buy these products? Is it the manager class that the capital class needs to keep the working class in line?
The fundamental challenge for the human species is to reimagine people not as economic entities, not simply as workers, consumers, owners but as beings of intrinsic worth alongside all other species. Capitalism has lead, and will continue to lead, to exploitation of all kinds (the global environment, people everywhere) not because it is evil but because it’s the way the system is designed.
As the Second Thought Youtube channel asks: why is it so difficult to imagine a life beyond capitalism and the increasingly unequal distribution of wealth? With the continual rise in AI prowess we urgently need an answer to this question and to imagine a future in which *everyone* can benefit from AI.
