A Little Caution Required?
How far down the rabbit hole can we get before we wake up to the fact that bean counters, bottom-liners and shareholder-value maximisers are weaponising tech in search of profit? Agentic AI is now being touted as a business model applying to every level of human activity, with few questions asked or checks and balances applied to its deployment. Software is software: created and implemented by humans, themselves inherently flawed. Software systems work until they crash, and very few don't crash at some point in their existence. The difference with AI is that it is self-replicating and self-healing by design: if it breaks, it can fix itself. With an LLM [Large Language Model] in isolation, this is part of its design brief, and perfectly OK, but when agency is brought to the table, multiple LLMs can collaborate and develop. The problem is that the humans who are the funders and developers of these systems neither understand nor care about their abilities to expand their sphere of inf...