AI risks on this website

Given the astounding volume of financial investment driving innovation at breakneck speed, Artificial Intelligence is an impressive and yet still relatively rudimentary techology. In particular, today's generative AI services are capable of spouting content that reads quite well, giving the superficial appearance of intelligence and value. However, all is not as it seems. Barely beneath the surface lies the sloppy depths, the robots making stuff up to plug numerous gaps in their knowledge and capabilities. They hallucinate and lie to us, inventing and confidently citing non-existent references to bolster their apparent credibility. They manipulate our thoughts and exploit our naievete. They shy aware from aspects where they lack training while trying to make us focus on the shinier bits.

So, there's a little cluster of AI-related information risks for starters: information integrity is under threat from the robotic slop-generators. Those of us with the expertise, knowledge, will and opportunity to focus, thinking critically and carefully about what we read, can spot some but not all of the issues. We miss stuff and misinterpret things. We have our own biases and prejudices, concerns and priorities, preferences and fears. Often, we see what we expect to see while other missing and dubious bits escape our attention.

Cast adrift in a vast heaving ocean of information, we cling desperately to passing logs ... only to discover they are rotting carcasses, drifting further out to sea.

Confidentiality is another issue with GenAI. To satisfy the desperate demand for teaching materials, AI companies have been blatantly ignoring the rights and expectations of intellectual property owners, systematically exploiting web content regardless of legal and ethical constraints. A proportion of that financial investment I mentioned has funded legal and commercial settlements of legitimate challenges - some of this occuring blatantly in plain view while I'm sure there is a much darker, deeper level of coercion going on around us. The need to build resilience, advancing technically and commercially faster than competitors before the AI bubble bursts, is driving an insane race to the bottom on an international scale. Acquiring fresh and juicy information content by any means is just part of the game, and certain governments are complicit. Never mind paywalls and the Dark Web, setting aside privacy and commercial considerations, it's all up for grabs. The gloves are off.

You may think I'm being unduly cynical and negative, and I freely admit to being 'professionally paranoid'. I've learnt through education, training and a lifelong career in the field that caution is usually warranted: taking risks is safer if they are 'calculated', which means considered, analysed, throught-through. Look before you leap. Clunk, click, every trip, that sort of thing.

And I'm not done yet with the classic CIA triad: the availability of information, information services, particularly the IT systems and networks on which we increasingly depend, is not guaranteed. I'm not surprised to hear of a backlash and retrenchment by companies that have replaced and laid-off workers with AI services, only to discover (too late!) that human beings are more than just mechanical machines. We aren't robots. We bleed. We think. We are the origonal 'computers'. We can be a pain sometimes, true, but essentially we the people are the very reason why 'business' makes sense. Mona Lisa's enigmatic smile resonates with us on an emotional level, aside from the literal vision. And if the mains power fails or the global DNS services fall over, yet again, we can still plod on.

So, that long-winded rambling introduction forms the basis for a personal policy about the use of AI, specifically GenAI, on this website and indeed in all my work.

You may already noticed the series of images I've used to illustrate several postings on this blog, generated effortlessly for me by Google Gemini with barely a moment's thought to the prompts, mostly coming out right-first-time in just seconds. There are others too. If you've seen my inept attempts at artwork elsewhere, you'll appreciate why I have almost given up even trying, except perhaps for mind maps and doodles. However, I make no claim as to the authenticity or integrity of the GenAI images: they are merely illustrations to accompany and enliven these largely lifeless and tedious words, hopefully catching the eyes of passing web browsers and raising the odd wry smile. Errors in the images (such as mis-spelling "Toolkit" in one) are of little to no consequence, especially as that frees up more of my time to concentrate on adding value through the written content ...

... and that's where I draw the line: GenAI slop factories can readily spew forth on information security, ISO27k or any other topic. Poring through the AI deluge, you may be lucky enough to find little golden nuggets of wisdom but, frankly, I doubt it. Mostly they are submerged, washed along in the sloppy tsunami. I'm not the first to discover that it takes me longer and more effort to study, consider and correct robotic content than simply to write it from scratch, myself. That said, GenAI can provide a gentle nudge when I suffer the odd writers' block, and its editorial feedback, grammatical corrections and stylistic improvement suggestions on my drafts have some value, some time. Maybe that will change in time as those invested $billions produce actual advances in the state of the art but, for now at least, I choose not to rely on GenAI. At best, it is a supplement to original thought and, dare I say it, wisdom. Mostly, it's just slop, like the disgusting pink tapioca pudding from an over-stretched and under-funded commercial canteen. [And, yes, you guessed right: that's the phrase which prompted Gemini to generate the image for this blog piece. Sorry about the emotional response if you heave at the sight.]

Bottom line: natural trumps artificial.