Tools like ChatGPT are easily weaponised, and the antidote is diverse human contact

Martin Geddes

I have held off from writing about AI for some time, especially ChatGPT, which gives a conversational interface to an automaton that impersonates an eloquent human via text. I have played with it, and waited to see the analysis of other commentators, before coming up with my synthesis. It has already proven to have utility to me in doing research on legal work, and its potential as “the next Google search” hardly needs me to chime in.

I have, however, “joined the dots” and spotted a gap in the insight market. Let me tell you the punchline right now: as currently constructed, AI like search and chat is optimised for the spread of mass consciousness, and works in opposition to the sustenance of political dissent, free thought, and a plurality of beliefs. That is because it is optimised to promote propaganda as fact, located in the mass consciousness of the collective. That is what its powerful funders (consciously or not) desire, so it is what gets built.

Let’s start by acknowledging that even an “genius generative plagiarist” is a technical tour de force. I have no idea if this is an original recipe, but this has potential for kick starting a beach party:

READ MORE