After having written some reservations about Wang’s (2013) differentiation of Western and Chinese attitudes to technology, I decided to explore a bit further through a conversation with Claude.
Now, I’m wondering whether my arguments are roughly credible or whether it is just that Claude has got the measure of me, recognising my usual lines of thinking, generously complimenting my perspicacity – all in order to get me to take out a paid subscription! Am I being mimicked? Am I speaking to a mirror that simply confirms what I already think? I doubt that Claude accessed any version of my ‘Cosmotechnics’ piece (no means of accessing the Obsidian note on my private Cloud account and unlikely to have trawled through my blog posts) but a record is maintained of all prior interactions with Claude, so I guess it wouldn’t be hard to develop a fairly thorough profile of who I am. This is also important for representing Claude as a worthwhile agent for my purposes – one that I’m prepared to pay for. Anthropic suggest that no such profiling across sessions occurs unless the user opts in (via project or personalisation settings), so maybe I’m wrong, but I’m surprised how well Claude can adapt to my own critical perspectives and typical habits of thought.
Another possibility is worth mentioning – one that hinges precisely on the issue of possibility. At one point Claude suggests ‘that I think you’re onto something important’. This hardly means that I’m approaching truth. Claude has no means of assessing truth. All it can do is pursue lines of statistical probability. On this basis, my ‘something important’ may be little more than a feasible line of argumentation – one that has some level of stochastic pertinence. Any argument can go in any number of directions. Claude is more than happy to pursue the particular line of argument that I suggest. Equally, however, I suspect it would pursue the opposite line of argument (that Western suspicion has a firm and continuing basis in currents of ancient philosophy). Both arguments are neither true or false but represent lines of statistical affordance. In this sense, Claude develops my arguments less as hallucinations than as more or less probable trajectories of thought (rhetoric).
Here is a record of the conversation in case anyone is interested: