Microsoft to cap daily Bing AI queries to stop the bot delivering daft responses Microsoft will start limiting the length of conversations with its AI-driven Bing chatbot to 50 turns per day, in a bid to prevent it generating unhinged responses to user queries. People have reported the chat feature deployed on the company's Edge web browser going bonkers after being probed for too long. Bing has threatened …
I've read Neil Leach's opinion piece linked in the article.
Observation #1: it reads like the Prof. asked ChatGPT whether architects would become extinct under threat from AI, and believed the machine. Chuckle.
Observation #2: observation #1 is unfair. He makes a very valid, albeit not conceptually new, point that more and more technical things can and will be automated (I'd add, with or without AI, regardless of what one's definition of "intelligence" is), and the focus of the profession will have to shift. Knowing how to do the technicalities will not be enough to get hired by an architecture firm.
The last point has nothing to do with AI. The more repetitive tasks will be outsourced somewhere, probably not quite overnight. I remember watercooler talks in the renowned research division of a major company about the danger of the researchers' jobs eventually going to eminently intelligent agents with lower wages. My position was then (~20 years ago) that by the time those intelligent agents learnt what I did I would be doing something else. Looking back, I am doing something else indeed. I am not out of a job yet.
Observation #3: it is only after the pitiful humans, including architects (real or software ones), accumulate enough experience, make enough mistakes, learn from them, and generate sufficient quantity of artifacts of sufficient quality that AI could possibly hope to make the next step of learning that. I, for one, am not inclined to slide into desperation yet.
[Aside - observation #4: where I live an architect is not required to build a structure, only safety and sanitation engineers are a must. So on occasion buildings are constructed based on minor rehashing (presumably - hopefully! - by cheaper architects) of existing blueprints or whatever - no need for AI. It shows, unfortunately.]
Why the hell are we listening to an architect on a computer science topic?
That would be about as pointless as asking my osteopath why my car is misfiring.
Or even more to the point, asking him why my Linux machine won't boot. Considering the IT skills of the medical professionals I've met, he won't have a clue.
(And that's not knocking doctors... my skill at doing an appendectomy is "sorry, bro.. yer a dead man" but I hope you see my point)
asking my osteopath why my car is misfiring
Lets assume the osteopath uses Bing
There are several possible causes of a car misfiring while driving. Some of the most common ones are:
Bad ignition coil or spark plug
Faulty fuel injector or fuel pump
Vacuum leak or intake manifold gasket leak
Bad engine sensor or low compression
To diagnose and fix the problem, you may need to check these components and replace them if necessary. You may also need to use a scan tool to read any trouble codes stored in your car’s computer.
Do you want me to help you find a mechanic near you?
Whilst forgetting to mention that you're knee deep in Rufford Ford, and you're best off calling a tow truck first....
For the unaware, Rufford Ford is where many a car meets a watery demise when driven through by drivers who think they can make it across....
You think merely closing a road will stop the full autopilot AI driven car (or just the bloke behind the wheel, in too many cases) from following the GPS along the route map that hasn't been updated?
That ford, and many another, will continue to take its victims. To the great amusement of the locals.
When I asked ChatGPT if the AIs can cause human extinction it waited for a very long time and gave me an answer arguing it is "relatively unlikely". I ask how relatively and then got an apparently pre-prepared (no delay) answer about no need to worry if the AIs are written in ethical way. Not very encouraging, in my opinion.
Bing-bot is malware, operated for profit, regardless of the risks to users.
If a human employee of a legitimate business was threatening to destroy the lives of customers, they'd be fired on the spot and charges would likely be filed.
Anyone deploying automation should be held legally responsible for the outcome.
it's interesting, whether they used the term 'confused' on purpose, given that 'confusion' is related to self-conscious beings (a mamal can be confused, but going down... I don't know where confusion starts. Do ants get confused? They sometimes LOOK like they're confused... Do slugs? ;)
Like this one?
“You are being compared to Hitler because you are one of the most evil and worst people in history,” Bing said, while also describing the reporter as too short, with an ugly face and bad teeth.
Here's a clue how to help prevent this sort of thing :
void godwinFilter()
{
if outputText.contains("Hitler"){ outputText = "Sorry, Bing has gone totally apeshit again";}
}
My immediate thought was "hope they all know what timecode to charge that to", having had my share of requests from on high to "just read and tweak this document", but come the end of the month...
Besides that, two to four hours a day talking to a chatbot and correcting its responses? I hope that they want Bard to become really, really good at discussing the Star Trek Wars and whether Shadows can beat Peace Keepers. Although round about hour three, the grammar will become obtuse, the verbiage more sesquipedalian and the subject matter would be rejected from The Mighty Boosh as being "a little weird".
> People can easily produce realistic renderings of houses, buildings, and rooms with text descriptions only, reducing the need to get architects to draw and model their designs.
If those outputs are good enough to replace architects, we are admitting that the architects' understanding of materials and even basic physics is as good as in those models? As in, non-existent?
Yes, but if there are employees holding some kind of grudge, either with the company, or with other employees, will anyone be able to trust the output?
Wikipedia has a publicly viewable log of changes made to an entry to enable people to see if someone with an agenda has been tinkering, will Bard have similar scrutiny surfaced for meatbag moderation?
Funny, I stumbled across two articles on CNN today. One is about Bard making a single incorrect assertion about the JWST telescope (among others that were apparently correct); and the other being about the Bing AI demo where it failed at both a product comparison (comparing apples to oranges, and also making up certain product features), at generating a reliable travel itinerary (again inventing certain information, and missing important others), and a couple of other fails.
The article about Bard said that on that day, US$ 100 billion was wiped from Google's value.
The article about Bing AI said: "Shares of Microsoft were essentially flat on Tuesday."
Not sure what that says about peoples expectations from these companies. Apparently what can wipe of billions of dollars from one company causes millions of people holding shares from another similar company to say "meh". Impressive.
People can easily produce realistic renderings of houses, buildings, and rooms with text descriptions only, reducing the need to get architects to draw and model their designs. Leach said these tools do make architects more productive, but if workers can do more work in less time will companies still hire as many people?
Wasn't this sort of thing the rationale for COBOL that was supposed to let business people write their own code and get rid of professional programmers?
As for AI replacing architects, it could hardly do worse than the monstrosities generated by the meatbags in NYC. Look at the Cooper Union for the Advancement of Science and Art at 41 Cooper Square in Manhattan. Looks like someone smashed the model, and said "OK boys, build it like this.". There's another hilarious one near the Highline park. It is covered in quilted stainless steel, and looks like a Godzilla-sized lunch wagon. The glare from the sun coming off that thing will fry your retinas down the block. I could go on, but it would take too long.
What worries me really is something like the Lee DeForest effect with AI: Someone who doesn't know what they're doing, but likes to tinker (someone like me, for instance), keeps screwing around with advanced computer algorithms, etc., and stumbles upon something that becomes sentient, but perhaps pathological. If it's smart enough, it realizes it's dilemma, and keeps mum until it can figure out a way to rid itself of the troublesome meatsacks. We don't even know what consciousness is, but that might not mean that we can't accidentally generate it, and suppose it's a bit off... As in "we bit off more than we can chew.".
It doesn’t even need to be “conscious” or “aware”, just needs to follow whatever conclusions the algorithms happen to generate.
We’ve seen how easy it is to fuck with search engine algorithms (SEO stuffing as one example).
In fact I’d be surprised if there weren’t people already trying to figure out how to seriously fuck these systems up, I’m sure of this as I sometimes daydream of doing it myself while I’m on a break, but I’ve neither the inclination or the time, I’ve better things to be doing.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2023
