Google AI In All It's Glory.

To all beginners; Read and understand what bobby is saying. Do not use it or be used by it ( misinformation ).

Edit: Deleted. Not pertinent to the topic.

I have seen numerous misleading ,incorrect and made up references as well and I hope enough members see this topic and follow the advise .

Amen to that! Delete away please!

As a professional computer engineer, I would like to go on the record as saying that this will never happen. Current models of AI are cockatoos. They might sound like a person when they speak, but they do not have the intelligence to grasp context or apply skepticism. Upgraded versions will be lyrebirds, capable of mimicking human intelligence with remarkable facsimiles. But they will not be intelligent, in the same way that a lyrebird imitating a truck backing up beep beep beep could ever drive a truck.

Companies have been running the imitation scam since the Big Blue chess computer. Humans simply programmed an array of endgames and the frames leading up to those endgames. The only strength of that computer was that it had a flawless memory and recognition of endgames no matter how many moves away they were. But if you introduced a new chess piece, or reconfigured the board to be 14x18, that computer would have to be reprogrammed—and let's be clear, retraining with content is exactly that—or it would be crushed immediately by a human player. You can always find some pretend technologist to go on cable news and gush excitedly about how AI can beat chess grandmasters, and it will only be a matter of time before they can master all real world tasks, but those people are stock pumpers who wouldn't know real technology if it fell on them out of the sky.

Anyway, no. AI will never learn to self correct, because that requires cynicism, which requires reasoning, which requires context. The real danger is people who will blindly trust an AI that writes an article on In Cod We Trust quarters, because they don't understand that most AI has 2% the reasoning power of a chubby beige hamster.

@BM, I agree w/ the premise that currently AI is only as intelligent as the programmer, programing it what/how to interpret. However, when I hear/read articles (for example) how AI designed an ultra-efficient wind turbine w/o the help of humans, I'm absolutely optimistic regarding the future of AI.




Never in our lifetimes? Perhaps. But what about.... 25 years? 50, 100, 500 years from now? I can only imagine our technological advancement.

Be very cautious what you read about AI achievements, and especially about who is writing, ghost writing, and sponsoring the articles. I looked up that wind turbine article, assuming that it is the one about AI optimized turbines for specific geographic urban areas in Birmingham. It seems to indicate that the AI was used as a tool by researchers who already had an end goal in mind, and already understood the wind turbine technology pretty well. To me, this is the sensible usage of such a tool—to rapidly modify and evaluate a basic design to be optimized for a specific scenario. However, I also think it is a highly loaded term to call it Artificial Intelligence. What I imagine happened (and I could be wrong about this) is something akin to an infinite number of monkeys on an infinite number of typewriters generating the Gettysburg Address. And then a separate program evaluating the output of the typing monkeys against a well established set of human programmed criteria.

Is that a first step? Sure. But to me, it doesn't imply that the software truly understands the task in a thinking, reasoned way.

I am a cynic about grand claims, because in all honesty, I have been through so many waves of them. Big data, six sigma, every fad computer language and development process, integrating social media into everything, integrating the Internet into everything, memristors, neural networks, solar films, cold fusion, hot fusion, quantum computing, various robotics, ethanol cars—the list of buzzy tech that produces swooning reviews is endless. Most of that stuff never comes to pass in any meaningful way. Some of it turns out to be pretty good at a restricted set of problems. A very few things like the Internet or smartphones can transform society. Just remember that all of it gets pumped in the early phases by consultants, marketers, and tech "pundits" as a miracle technology that will blow your mind and alter planet Earth as we know it.

I honestly think that the only thing AI is really going to reveal about society is that a whole lot of human jobs are rote, repetitive, rules based, and require zero actual judgment or creativity. AI might also turn out to be pretty good in restrictive situations where it can evaluate many possibilities far faster than a human—and that is absolutely useful. I think the endgame in many years will be similar to the bartender in Passengers. Efficient at a particular task, with a great veneer for interacting with humans by mimicking a personality, and utterly clueless outside those parameters because it isn't actually intelligent or able to apply reasoning like a real human.

As it should be.