The blurb. From my head.
Okay, the number in the title is made up. It’s probably much larger than this.
The thing is, I keep seeing this image.
With loads of percentage stats attached (always different). And yet not one person has questioned where the data has come from, or what the baseline was before the percentages were applied. And 3.2m x 2,500 isn’t even 8.1bn.
And many people are concerned that critical thinking skills are being lost. Might already be too late for the LinkedIn users. And calculators.
Okay, so strikethrough text in a headline doesn’t appear to be as easy as it should be, but really that’s not the point of this post.
Today I read about a PwC website for the newly minted ai training much trumpeted by the UK government and lambasted by just about everyone else. Turns out the website cost a whopping £4.1m to make. And all it does is list links to places where you can get some training. Mahad Kalam has all the info. Basically, a bookmarking site.
Then I caught a thread on BlueSky that detailed what it’s like to try and use it. I think the term is Epic Fail. (Some people noted that this wasn’t able to be read as they are blocked by the author due to following someone who isn’t liked by the author of that thread, so here’s a post that collects the thread into a single web page you can read.)
Turns out it’s not money well spent, it’s just money spent trying to show how amazing ai is and not only failing to do so, also making a mockery of government procurement, internal testing, and testing at any point with its core audience. So in some ways it’s a success. Er, right?
I’ve noticed more and more that LinkedIn posts about ai come in two very distinct flavours:
Now, I probably fall into the latter for most things, and there’s a reason for this. ai is a lens. It looks at tasks, say summarising a document, through a lens. It’s not a lens I know or can know, and therefore I am less trustful of it’s summary. I can read it myself and get a summary in a reasonably short time and I also get the context around that summary. I can know it.
Where ai (by which I mean machine learning) is great is where you’re needing to find the equivalent of a needle in a haystack and to know if it’s hay or the needle. It’s brute force summarising, as it were. I don’t care how it’s done, it’ll find the needle and present it to me.
Sometimes I need both. I need to find the needle quickly, but also need to know that I can find the needle myself – or better, understand how the needle was found – as I can build out context from this. An example would be video call transriptions. I don’t find them helpful on their own, but with notes I make about body language and cadence, etc, I can build a really interesting picture about what’s important. ai, as yet, cannot. I suspect it won’t. Certainly in my lifetime. Because it’s good at using the past to determine its output and not very good at understanding the present and all its nuances and then using these to add colour.
It’s quantitative not qualitative.
Oh, and the reason I don’t want to use ai in products and actively turn it off? They’re using it to feed their machine, even when they say they aren’t. I can’t trust them. I mean, it’s not like tech companies have proven themselves to be trusted, right?
I’ve decided, to make sure I don’t over-think these posts, to type my thoughts straight into the interface and publish when I’m done. This isn’t because I think what I have to say is right or complete, it’s because I want to use this space as a place to think out loud. I actually want to be wrong occasionally, or at least to have be happy to be wrong later on when I’ve had more time to think things through properly.
So, that’s the premise of all these posts. I need to let go the things that are whirling around in my head – and have been for some time, if I’m honest with myself.
—
I keep seeing people talking about ai as a way to speed up their thinking. I’m not convinced that’s what it can do, but I’m definitely certain that it shouldn’t be used in that way. Rather than use ai (always lower case) to speed towards making a decision on something, I’m more interested in using ai to enable humans to be able to slow down enough to think critically about what they are doing, will do, and have done.