It's probably not the recommended course of action going from a melanoma diagnosis to the beach, but here I am. Jane has a conference at Broadbeach today so I drove down with her and I'll be working in this apartment with a nice view of the water for the next day or so. Probs not out on the deck, though.
I should preface this comment with the disclaimer that I work in the advanced analytics department of my employer, and we're using AI and other tools to optimise our operations.. That said, I rely on the actual smart people (data scientists, engineers and analysts) to do the work, I'm just a business analyst dealing, ironically, with the human interaction and expertise bit.
I’m with you Andrew in that it’s not really “intelligence”, artificial or otherwise. Machine learning is more accurate but still not quite there. The uses we’re finding for it is in crunching enormous data sets and finding patterns and insights it would take a human days or weeks to manage on their own, if they could manage it at all, and without the, er, risk of human error.
The new washing machine I bought last year apparently has “AI” to help me optimise my loads. What it really has is an algorithm that detects the load cycles I use the most (bedding, towels, intense cold) and presents them to me when I turn the machine on. Which is useful but hardly “intelligent”.
The way everything now has “AI” in it tells you what a gimmick most of it is. Working with AI tools it's clear that for all their potential the actually valuable aspects of them do not exist without human input or intervention; in my work we need human oversight AND insight - human expertise - to render value from what the AI is producing. The most valuable things our AI is doing is either speeding up tedious manual work, for example extracting information from product info sheets automagically that would otherwise require a human being to trawl through and manually collate, or speeding up/enhancing human decision making or analysis - crunching data that humans did not have either the time or the capacity to do themselves, and then using that to make decisions. While we have unlocked insights and efficiencies with AI that were previously not accessible, we are nowhere near replacing the human element in the process. So much of what we're doing is tempered by the human experts going "yes that's very interesting but it doesn't take into account A, B, C because of X, Y, Z..." and then we have to go refine our models and business rules again. And that's just for the bits where we can identify patterns and data; there's so much that's essentially ineffable, driven by human instinct, observation and experience that AI models can't begin to replicate.
I would argue that we won’t have genuine artificial intelligence until we have artificial sentience; and that’s likely to be decades if not centuries away. Space Karen can argue that AI will be smarter than us in a few years but he’s full of it. What we’ll have is a highly refined algorithm that can collate and parse data at a speed and complexity never before seen, and therefore process that data in ways we hadn’t imagined were possible until now, but that doesn’t make it intelligent.
That said, I'm all here for tools that help you do your best more efficiently or better; how much of human productivity is stymied by having to arrange and collate your notes/thoughts/stream of consciousness into something usable, and by then the idea has floated away never to be revisited?
" ramble into your microphone of choice for up to 10 minutes and then spits out a really well organised document" well that just sounds like a challenge. I want to read it 10 minutes from Burrough's Naked Lunch or Joyce's Finnagan's wake and lets see how well organised a document it produces.
I also have seen some using the term Machine Learning instead of AI. I expect its too late to see it become widely adopted but I think its a more useful description with less inherent hype than the term AI.
I should preface this comment with the disclaimer that I work in the advanced analytics department of my employer, and we're using AI and other tools to optimise our operations.. That said, I rely on the actual smart people (data scientists, engineers and analysts) to do the work, I'm just a business analyst dealing, ironically, with the human interaction and expertise bit.
I’m with you Andrew in that it’s not really “intelligence”, artificial or otherwise. Machine learning is more accurate but still not quite there. The uses we’re finding for it is in crunching enormous data sets and finding patterns and insights it would take a human days or weeks to manage on their own, if they could manage it at all, and without the, er, risk of human error.
The new washing machine I bought last year apparently has “AI” to help me optimise my loads. What it really has is an algorithm that detects the load cycles I use the most (bedding, towels, intense cold) and presents them to me when I turn the machine on. Which is useful but hardly “intelligent”.
The way everything now has “AI” in it tells you what a gimmick most of it is. Working with AI tools it's clear that for all their potential the actually valuable aspects of them do not exist without human input or intervention; in my work we need human oversight AND insight - human expertise - to render value from what the AI is producing. The most valuable things our AI is doing is either speeding up tedious manual work, for example extracting information from product info sheets automagically that would otherwise require a human being to trawl through and manually collate, or speeding up/enhancing human decision making or analysis - crunching data that humans did not have either the time or the capacity to do themselves, and then using that to make decisions. While we have unlocked insights and efficiencies with AI that were previously not accessible, we are nowhere near replacing the human element in the process. So much of what we're doing is tempered by the human experts going "yes that's very interesting but it doesn't take into account A, B, C because of X, Y, Z..." and then we have to go refine our models and business rules again. And that's just for the bits where we can identify patterns and data; there's so much that's essentially ineffable, driven by human instinct, observation and experience that AI models can't begin to replicate.
I would argue that we won’t have genuine artificial intelligence until we have artificial sentience; and that’s likely to be decades if not centuries away. Space Karen can argue that AI will be smarter than us in a few years but he’s full of it. What we’ll have is a highly refined algorithm that can collate and parse data at a speed and complexity never before seen, and therefore process that data in ways we hadn’t imagined were possible until now, but that doesn’t make it intelligent.
That said, I'm all here for tools that help you do your best more efficiently or better; how much of human productivity is stymied by having to arrange and collate your notes/thoughts/stream of consciousness into something usable, and by then the idea has floated away never to be revisited?
I can test on my dad if you like. It might cause the machine to leave the room.
" ramble into your microphone of choice for up to 10 minutes and then spits out a really well organised document" well that just sounds like a challenge. I want to read it 10 minutes from Burrough's Naked Lunch or Joyce's Finnagan's wake and lets see how well organised a document it produces.
I also have seen some using the term Machine Learning instead of AI. I expect its too late to see it become widely adopted but I think its a more useful description with less inherent hype than the term AI.
Ooooh I want me some of that word salad organiser