Science Fiction
Dictionary
A B C D E F G H I J K L M N O P Q R S T U V W X Y Z
 |
Is Agentic AI The Wrong Kind Of Smartness?
In a recent TED talk, former Google CEO Eric Schmidt is talking about what happens when AIs start moving from problem to solution in ways that we can't understand or figure out.
[8:00] Now we're all busy working, and all of a sudden, one of you decides it's much more efficient not to use human language, but we'll invent our own computer language.Now you and I are sitting here, watching all of this, and we're saying, like, what do we do now?
The correct answer is unplug you, right?
Because we're not going to know, we're just not going to know what you're up to. And you might actually be doing something really bad or really amazing. We want to be able to watch.
So we need provenance, something you and I have talked about, but we also need to be able to observe it. To me, that's a core requirement.
There's a set of criteria that the industry believes are points where you want to, metaphorically, unplug it.
One is where you get recursive self-improvement, which you can't control. Recursive self-improvement is where the computer is off learning, and you don't know what it's learning.
That can obviously lead to bad outcomes.
Isaac Asimov talked about this in his whimsical 1975 story Point of View. In this excerpt, a computer scientist is talking with his son about Multivac, whether or not computers need time to play and the problem of how to figure out if the machine is wrong about a complex problem:
"And the thing is, son, how do we know we always catch Multivac? How do we know that some of the wrong answers don’t get past us? We may rely on some answer and do something that may turn out disastrously five years from now. Something’s wrong inside Multivac and we can’t find out what. And whatever is wrong is getting worse.”"Why should it be getting worse?” asked Roger.
His father had finished his hamburger and was eating the french fries one by one. "My feeling is, son,” he said, thoughtfully, "that we’ve made Multivac the wrong smartness.”
"Huh?”
"You see, Roger, if Multivac were as smart as a man, we could talk to it and find out what was wrong no matter how complicated it was. If it were as dumb as a machine, it would go wrong in simple ways that we could catch easily. The trouble is, it’s halfsmart, like an idiot. It’s smart enough to go wrong in very complicated ways, but not smart enough to help us find out what’s wrong. — And that’s the wrong smartness.”
On the subject of computers (or robots) developing their own language and talking just with each other, see Robot-Control Wave Band from Rex (1934) by Harl Vincent, mentanical communication from The Mentanicals (1934) by Francis Flagg and Autonomous Car Intercommunication from Sally (1953) by Isaac Asimov.
Scroll down for more stories in the same category. (Story submitted 6/1/2025)
Follow this kind of news @Technovelgy.| Email | RSS | Blog It | Stumble | del.icio.us | Digg | Reddit |
Would
you like to contribute a story tip?
It's easy:
Get the URL of the story, and the related sf author, and add
it here.
Comment/Join discussion ( 0 )
Related News Stories - (" Computer ")
Cortex 1 - Today A Warehouse, Tomorrow A Calculator Planet
'There were cubic miles of it, and it glistened like a silvery Christmas tree...' - Clifford Simak, 1949.
Is Agentic AI The Wrong Kind Of Smartness?
'It’s smart enough to go wrong in very complicated ways, but not smart enough to help us find out what’s wrong.' - Isaac Asimov, 1975.
Jetson Orin Nano Super 70 Just $249
'Rayno folded up the microterm and tucked it back inside his jumper.' - Bruce Bethke, 1983.
Automatic Bot Traffic Is 38 Percent Of HTTP Requests
'there were so many worms and counterworms loose in the data-net...' - John Brunner, 1975
Technovelgy (that's tech-novel-gee!) is devoted to the creative science inventions and ideas of sf authors. Look for the Invention Category that interests you, the Glossary, the Invention Timeline, or see what's New.
Science Fiction
Timeline
1600-1899
1900-1939
1940's 1950's
1960's 1970's
1980's 1990's
2000's 2010's
Current News
FTC: Says Ring Employees Illegally Surveilled Customers
'Then she looked up with a smile and moved closer to the camera.'
Switzerland May Cap Population At Ten Million
'The population of Castle Hagedorn was fixed...'
Project Silica Offers 'Long-Term' Digital Storage
'... folios and tapes and playable discs of platinum alloy.'
Can 'Tactical Umbrellas' Shield One From Drones
'... another corner of his mind began to think about the shields.'
Crystalline Structures In Space, You Say?
A massive space borne lifeform from ST:TNG.
Garçon! A Menu For Artemis II, S'il Vous Plaît
'Michel Ardan, as a Frenchman, was declared chief cook, an important function, which raised no rival.'
Amazing Photonic Crystal Light Sail
'That sail will be twenty thousand miles at the wide part.'
Blue Collar AI Goes To Work To Mine Its Own Crypto
Blue collar bot.
Rogue AI Replicated Itself
'Sapiro’s computer just kept dialing at random, hanging up on humans, until it got a fellow computer of the same type as itself.'
HandelBot Helps Two-Handed Robots Learn Piano
'I request that you feed the correlation between those dots and the levers of the panel into my memory banks.'
Woven Fiber Electronic Skin For Robots
'... all the feel and appearance of human flesh and epidermis.'
When AI Takes Its First Breath
Any suggestions?
Chinese Aircar Light And Airy, Not For Blade Runners
Daytime version.
The Morphing Wheel And The Smartwheel
'If you surf over a bump, the spokes contract to roll over it.'
Transporting Antimatter
'...drawing plans for the magnetic tongs and bed plates and relays.'
Polish Turns Your Nail Into A Stylus
'He wrote on it, using the pointed fingernail of his right forefinger...'
