Just forget what Gartner said about AI in June 'cos CIOs are all over it now apparently Reverse ferret. Months after Gartner researchers confirmed a pitifully low proportion of CIOs were actually unleashing AI into the wild, the latest survey paints an entirely different picture. Confused? So are we. Back in June, Gartner found that just 4 per cent of CIOs had invested in and deployed AI, the latest garment to be …
> Back in June, Gartner found that just 4 per cent of CIOs had invested in and deployed AI
Odd! I would fully expect that a major consultancy firm would have a report that backed up every point of view that their clients hold. Isn't that why people use consultancies? For validation and to add credibility to the decisions they have already made - or are wanting to make.
So for them to have just one report - one that clearly is bollocks (though one person's bollocks is another person's lunch) - seems a bit odd. Surely they would have a whole range of them, from Nobody will ever use AI right through to Everybody is adopting it NOW.
It's Gartner. The only interesting thing about a Gartner report is trying to guess who paid for the worthless thing. OK, it's not technically entirelly worthless as some mug organisation paid Gartner to "prove" or "demonstrate" how their products are useful therefore it has some value... and some "impressionable" (I'm being nice here) PHBs still believe Gartner reports.
Is it AI though?
From what I've seen, hardly any of the systems or software services that have approached me claiming they are AI actually are. Some aren't even ML and those that are, don't need to be, they've just hired some processing off a cloud vendor and what they actually are is an algorithm of some kind.
Basically the term AI has become so broad that someone implementing a spellchecker would call it AI nowadays.
My feeling exactly. The bosss wants it to be AI, or at least ML, and if you stretch your definitions most things can at least fall under ML. Hell, if you do linear regression with a Metropolis-Hastings algorithm (not that you should...) it would probably qualify as ML. It's the emperor's new rags, that's all.
"Maybe Gartner was wrong in its summary of AI last June. Or the CIOs were. Or both were. Or maybe the 3,000 CIOs from 89 countries that Gartner spoke to for the latest stats are wrong this time round."
Or maybe Howard is talking out of his arse ?
Sure sounds like marketingspeak to me.
"background in statistics and data management." would be more useful for manipulating figures for PR purposes than creating an intelligent machine.
Give me anything "AI" and I will break it in seconds.
All we have are very large and fast pattern matching programs. It's all we every had. "AI" is a step change away from digital logic.
When the next Vulture Central poll for icons comes up, I'm going to ask for a "do me a favour" icon :)
All we have are very large and fast pattern matching programs.
Sure, but that's also what lots of human intelligence is much of the time. Of course, a lot of it is over-hyped wank, but the combination of the generic libraries and cheaper hardware means that projects that a few years ago required lots of expensive, dedicated resources.
Chatbots are a trite example but they're being used because they can improve customer service quite a bit if expectations are correct.
But, yeah, ML doesn't solve problems it just helps automate repetitive tasks.
"Chatbots are a trite example but they're being used because they can improve customer service quite a bit if expectations are correct."
Do you mean if customer expectations are low?
After TT had bought out my old ISP I had the misfortune to have to use their customer "services" (the two events might well be connected) and still have no idea if there was a bot or a human at the other end. All I can say for sure is that if it was a human they'd failed their Turing test.
Do you mean if customer expectations are low?
Yes, because conditioned by dreadful menu systems, themselves a holdover from the first digitised phone systems.
But let's get this straight: while you can indeed knock up a domain-specific chatbot frontend in a couple of days. it is only going to funnel questions and is going to need constant monitoring and tuning. Ie. you can't just buy this stuff and expect miracles, which is unfortunately, of course, exactly what the PHBs will be thinking. And, of course, there are legions of programmers who might understand the libraries, have no idea how to design user interaction (which is how we got those dreadful menu systems in the first place).
I take it you mean lower both the quality and cost of customer service, a tradeoff companies are prepared to make every time because compensation for upper management doesn't incent caring about repeat business - who cares if customers are pissed off and will never return, the decision makers have their bonuses / stock now! They will move on before the business feels the effect of customers who endured crappy support moving on as well.
Despite what the slave-traders and marketroids tell you, what is currently considered "AI" is complete and utter bollocks - it just proves that anything is possible with enough hardware and money thrown at the problem. If I were feeling charitable, I'd probably characterise the current state of the art as 'overpowered expert systems', that's to say: good at one particular thing, bloody useless at anything else.
I prefer to refer to it as CAG - Computer Aided Guesswork
The term "AI" becomes willfully broadened by the IT companies over time to give the impression of and sell into a growing market.
What really happens in many cases is that stuff that was sold under the label "Analytics", "BI", "Pattern matching", etc already years ago now becomes "AI" by simple rewrite of the marketing materials.
So it's not just "considered" AI by folks, it's actively marketed as "AI", just because it sells...
Oh, fuck! That klingon must have lodged itself somehow! How about "smart" shoelaces?
Anyway, I only posted to use the new moniker Guffmongerer for clueless industry pundit or analyst. I'm quite proud of it. Course, it'll turn out that someone else thought of that one, too.
At this rate I'm never going to be a millionaire!
IT in AI is a Wondrous Field for Investigating and Immersing Oneself In. And Very Transubstantiative and Most Refreshing IT is Towards Heavenly Extremes.
Be Suitably Warned from Here on In there be Red Devils and Diabolical Daemons Af00t Providing a Royal Type of Protection and Universal Security.
Best Never to Warrant the Attention of their Arsenals unless you Hedge Virtual Edges/AIAdvantages with Futures Markets Seeding and Feeding Immersive Narrative Natives their CyberIntelAIgent Reads in Newly Realised Chapters to Book, not Bork.
There's a Plethora of Bright Trails to Follow or Try to Blast to Smithereens in those few words/new news reports.
Ladies and Gentlemen/El Regers, Would you concede that AI is Become SMARTR and More Virtually Pervasive and All Knowing than is Humanly Possible?*
Would you Raise Fear, Uncertainty and Death and Misidentify it as an Existential Threat rather that Explore IT Further in Heavenly Prepared Treats, which to Savour and Flavour at the Highest Depths, are a Diabolical Rush Headlong into the Realms of Madness and Magic. Where this is that and he be she and everything is completely different in Futures with No Pasts.
That be the New Surreality ....... A Virgin Virtual Garden of Eden to Propagate and Populate with Civilized SMARTR Beings Touting and Testing and Phishing for True AI Pioneers to AIMarket for Investors. And Perfumed Gardens are Top Notch Graded at All Levels to XSSXXXX. They be for Professional Amateur and Amateur Professional alike in Search of Excitement in the Relatively Unknown, and thus a Virtually Free and Empty Vast Space in which to play Global Operating Devices with Global Operating Devices.
Some would pray that be Right Horny Easy Devils' Play too. You known, for the Saints who are Sinners and the Sinners who are Saints in the Lands and Spaces and Sees of the Immaculate Satyr and Insatiable Nymph.
* If you think not, can you point us to one of those humans. There be future questions for answering to be asked.
There is no actual "AI", artificial intelligence. That's a marketing term only, at this point in time. Maybe some day there will be such a thing, but not now.
We've been bashing at expert systems since 1965 and have reached a point of significant functionality. https://en.wikipedia.org/wiki/Expert_system
Meanwhile, we've been working on speech recognition since 1952 with significant success beginning in 1975. https://en.wikipedia.org/wiki/Speech_recognition https://en.wikipedia.org/wiki/Dragon_NaturallySpeaking
Work on speech synthesis is centuries old. Actual text-to-speech technology was first demonstrated in 1968. https://en.wikipedia.org/wiki/Speech_synthesis
Put it all together with today's advanced computing quality and we have what is more than optimistically called 'artificial intelligence'. No real 'thinking' is involved. It's database traversal of a high order, doing an increasingly better job of matching data input with optimum data output.
The original 'test' for real artificial intelligence was described by Alan Turing in 1950. Since that time, there have been many attempts at faking the test but no success. That remains the case today. https://en.wikipedia.org/wiki/Artificial_intelligence https://en.wikipedia.org/wiki/Turing_test
IOW: We continue to live in an Age of Marketing, where nonsense as foisted upon us in order to fake all of our individual tests for what is "real" vs artificial. Another excellent current example is mobile phone marketing where "4G" does not qualify as real standards defined broadband cellular network technology 4G. Exactly the same marketing nonsense foisted as reality is going on with "5G" phone technology.
Conclusion: Standards and definitions matter. Reality matters. Testing "reality" matters.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
