News: 1713119168

  ARM Give a man a fire and he's warm for a day, but set fire to him and he's warm for the rest of his life (Terry Pratchett, Jingo)

Why making pretend people with AGI is a waste of energy

(2024/04/14)


Interview While the likes of OpenAI and Google DeepMind chase after some fabled artificial general intelligence, not everyone thinks that's the best use of our time and energy in developing AI.

Computer scientist Binny Gill – CEO and co-founder of business automation firm Kognitos, and formerly chief architect and cloud CTO at Nutanix – thinks the push for AGI is the entirely wrong approach in what could be the next industrial revolution. Rather than trying to replicate humans with some kind of general-purpose artificial intelligence, Gill thinks we should look to the past to see what sort of systems we should be building.

"I believe that we need to learn from the industrial age a bit and see how we had offloaded manual labor, and then take those learnings and figure out how we can offload mental labor," Gill told The Register .

[1]

"We created a whole plethora of different looking machines. And they were not mimics of human arms and legs. Cars don't have legs. Elevators don't have arms to climb up ropes."

[2]

[3]Youtube Video

Gill instead hopes we'll see the rise of what he calls artificial narrow intelligence, or ANI. This isn't a new concept; it's the sort of application-specific machine learning that already exists behind things like self-driving cars.

[4]

And even if machines end up with intelligence and superhuman abilities, we still need to be in the loop. "It takes a human to decide eventually what is right and what is wrong," Gill told us. "You need the Iron Man inside the suit."

To learn more about Gill's optimistic vision for the future of AI, watch our full video interview with him by clicking on play above. ®

Get our [5]Tech Resources



[1] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=2&c=2ZhxR@qJmZXS48Gx63GUo@QAAAAA&t=ct%3Dns%26unitnum%3D2%26raptor%3Dcondor%26pos%3Dtop%26test%3D0

[2] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44ZhxR@qJmZXS48Gx63GUo@QAAAAA&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0

[3] https://www.youtube.com/watch?v=vPSVadJpbOM

[4] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44ZhxR@qJmZXS48Gx63GUo@QAAAAA&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0

[5] https://whitepapers.theregister.com/



kameko

What a reasonable thing to say. Too bad reasonable isn't what attracts investors.

He gives some funny examples on what actual useful machinery is, but I think my favorite example is the vacuum cleaner.

I mean, good lord, how many science fiction stories from the 50s to the 90s imagined we would have hulking mannequins marching around the house pushing around a regular old vacuum cleaner? And when we got actual robotic vacuum cleaners, what did they end up looking like? A little disc on the floor.

You know, I'm not even sure if I can call AI good at being creative, or if it's just that humans are absolutely horrible at being creative. Any time we ever try to think of something new or alien, we just think of ourselves.

This is why I absolutely adore HAL 9000's design. Did you know Arthur Clarke originally was just going to make HAL a boring old android? But no, he did something absolutely revolutionary for the time. Instead of making HAL a big, beautiful, noisy avatar to glorify the idea of an intelligent machine, he just... put him in the wall, like a thermostat. And that's insanely practical, it very much is like what smart home appliances are like today. That's what actually happens to technology and robots, they don't get bigger and better, they get smaller and more invisible. I don't want an oversized barbie doll running around my house pushing around a noisy vacuum, I want a little disc on the floor I barely notice. Did you know a Keurig is technically a robot? People don't even think of it like that, it's just that invisible and out of the way. It's kind of like how people commonly used to think computers would only get bigger, not smaller (that's why HAL's computer chamber is so big, which is one of the incorrect speculations 2001 makes). People absolutely suck at guessing what will be new, if not even think the opposite of what will really happen.

The sad fact is ...

Adair

some of us (money grubbers, the terminally narcissistic, and a bunch of other sociopaths), have never got over not living in a culture based on human slavery, and are constantly seeking to re-establish such.

It's all about power, dominance (and money, obviously). It may also be about a profound sense of insecurity.

Never forget this saying, it sums it up perfectly

ecofeco

"A future where A.I. creates art and writing and humans still do the dirty work is not the future we deserve."

Doctor Syntax

I'm convinced that real intelligence comes from the fact that we are physical entities interacting with the physical environment around us. We start out, newly born with millions of sensory receptors - rods and cones, taste buds, olfactory receptors, those cells in the cochlea, touch, heat, pain receptors, proprioceptors all feeding into a growing brain and spend the first months correlating those inputs to start building a mental model of the world around us, the perceived world. We have a sense of self stemming from our connection to the world.

As we refine it we can use it to understand that external world, plan and carry out our interactions with it to do and get what we need and want. Most o our vertebrate brethren do the same. That's the core of our intelligence.

What we humans have evolved is the extra capacity to communicate symbolically with language so that we can add to our model by listening or reading to others telling us about things they have encountered and we haven't and also to add more from our imagination and to reason more elaborately about our mental model. But that mental model of the environment is the core of our intelligence. The symbolic layer is just an add-on.

What ML and attempts at AGI are targetting is just that add-on layer. Without the perceived world underpinning it they're not going to get there. You and I can read a description of a room, a landscape, a person or experience, fictional or otherwise and add it to our mental model because we have seen and experienced these things directly during our training of our brains in those early years and beyond. A notional disembodied AGI can't have that connection to give a sense of self or an understanding of anything that's communicated to it. No amount of description can give it the sense of what a drink is because it's never felt the weight of the vessel and contents, never tasted it, never sensed the tempperature and never swallowed it. It is and must remain deficient in kind, not just in degree from a real intelligence.

Gunter's Airborne Discoveries:
(1) When you are served a meal aboard an aircraft,
the aircraft will encounter turbulence.
(2) The strength of the turbulence
is directly proportional to the temperature of your coffee.