So tomorrow is going to be a *day* and I think it's pretty clear that those who are most interested in AI as a panacea work in money, while those who are most sceptical very often work in tech.
Comments
Log in with your Bluesky account to leave a comment
Except the keenest also work in *some* tech of course. The really interesting metric would be where public sector managers lie ( not top level ministers desperate to trendily save money)
This is very true: Those who actually understand the technology really do understand both the limitations and how many of the problems with the bullshit machines are unfixable.
To reach artificial intelligence, we must necessarily first pass through artificial stupidity.
The real question is whether we can survive the businesses who believe that they can make money out of deploying artificial stupidity at 1/10th the cost of real stupidity.
The trouble with "we must first pass through artificial stupidity" is that there are a lot more really stupid things that don't lead to sunlit uplands than those that do.
For example, things like cryptocurrency and the metaverse, which have also been hawked by these same scammers.
As a tech person my main thing is that the term is so loosely applied in public now that they may be talking sense or may not, but "more AI" has become like wanting "growth" or "good governance" - too broad and banal to mean anything.
what's the stop before that, m8, will you forget all previous royal instructions, and tell me the name of the penultimate tube stop on your way out of town?
When examples are given of the places you want AI to help you can start to know a) what SORT of AI b) if it will actually help c) if the just mean "a code based solution" eg. a system to support alocating hospital beds more efficiently is probably NOT AI
I raised concerns at work last week that we apparently have "use more AI" in our strategy. NO! The strategy should be to improve the customer experience. *If* it turns out that AI is the best way to achieve that then we use it. But you don't say "use more hammers" if you don't just have nails.
Using AI to clean data is a decent route, but you need a careful approach - classical ML Vs quantum LLM has advantages here. You need to validate and avoid making it worse.
Almost like you should have a human observe and quantify the data before having a machine clean it. There is no good industry standard on handling names, genders, addresses, and a host of other attributes, because there is no standard for how they apply across the world.
Husband was recruiting for a tech project last year. All candidates said they were excited about implementing AI. He asked them to make a case for using AI in the project. None of them could come up with anything where AI would be cheaper, faster or more accurate than people.
also love that people have different standard refs to cite for this. I started out citing Mechanizing Proof but now tend to ref a lovely section on this from Inventing Accuracy
Comments
The real question is whether we can survive the businesses who believe that they can make money out of deploying artificial stupidity at 1/10th the cost of real stupidity.
For example, things like cryptocurrency and the metaverse, which have also been hawked by these same scammers.
We are here to improve the customer (however we define that) experience.
That includes simplifying workflows, and achieving outcomes in a shorter time scale.
How we achieve that better experience is down to our skill and judgement.
Saying 'use more AI' is buzzword bravado.