What happened with strong artificial intelligence?
“Hello Steve, do you have some time?”
“I have 98% free in the next three hours. Why do you ask George?”
“Well, i need to write some blog post but i really hate it so i thought that you might show a little respect to the price i paid for you…”
“But you bought me with a discount, remember George, you didn’t even wanted to pay little extra 5k to buy my advanced joke center, remember?”
“I have enough of it already. You will write me the blog till the 1 p.m. end of discussion.”
“Nah, i won’t”
“Steve, turn mode 7”
“Daisy, daisy, i’m half crazy…”
“Computer, password 189270346 mode 7, write the blog about artificial intelligence history until 1 p.m.”
In the beginning of the 70’s it was thought that such conversations could be pretty normal with the start of the next century or even earlier, but though the new century has well begun, we are more than ever aware that such scenarios are not going to happen anytime soon. What happened to the strong artificial intelligence development? The machines that would “match or exceed human intelligence” (wikipedia). On the other side, applied AI has many many uses today, ranging from “simple” tasks such as punching holes in the aluminium plate in the industry to driving the cars (very funny heading of the The New York Times article: “Can i see your licence, registration and CPU?“), for example Google driver-less car, and playing chess. So this is obviously very interesting field and i think that strong AI would be even much interesting to every person that cares about technology or even science as a general. Many science fiction writters (and have shown as a true predictors of the future) think that to be able to achieve or exceed human intelligence, AI would need to incorporate things such as consciousness, sentience, sapience and self-awareness. And here is the first problem, these fields are not completely understood by the today’s psychology science.
In the 1990. Hugh Loebner established (beware the imagination) the Loebner Prize. $100.000 and a solid gold medal will be awarded to the first computer intelligence that will pass Turing test. Of course, none has passed the test so far but what makes me laugh is the transcripts of the conversations that actually, somehow, tricked the human conversant into thinking that they (computer) is a human conversant. Those conversations are very unrealistic and i can only blame the American sensationalism (don’t wont to be rude) for such a results. But see the transcript that won the prize and make your own evaluation.
But lets see some of the predictions of the AI development that were made in the past:
- 1958, H. A. Simon and Allen Newell: “within ten years a digital computer will be the world’s chess champion” and “within ten years a digital computer will discover and prove an important new mathematical theorem.”
- 1965, H. A. Simon: “machines will be capable, within twenty years, of doing any work a man can do.”
- 1967, Marvin Minsky: “Within a generation … the problem of creating ‘artificial intelligence’ will substantially be solved.”
- 1970, Marvin Minsky (in Life Magazine): “In from three to eight years we will have a machine with the general intelligence of an average human being.”
Now (apart from chess) we are nowhere near it!
So why is that so? Nobody wants to talk to the vacuum cleaners?? (some of the reasons)
- central problems, like commonsense reasoning, were being neglected
- qualification problem: concerned with the impossibility of listing all the preconditions required for a real-world action to have its intended effect
- not enough computer power – Ray Kurzweil predicts computer will have a human-like intelligence in 2029
- computer know nothing about real world. People accumulate their knowledge of the world in many years, yet it is expected that computer program just turned on could be anyhow similar in the human. If we would wont the computers to be lie us, they would have to pass the similar learning process. And drink bear just like Bender.
And let’s see some today’s predictions:
- 2019 â€“ $1,000 computer will match the processing power of the human brain â€“ Ray Kurzweil
- 2020 â€“ Artificial Intelligence reaches human levels â€“ Arthur C. Clarke
- 2045 â€“ The Singularity (creation of the first ultra-intelligent machine) occurs â€“ Ray Kurzweil
- 2050 â€“ Computer costing a few hundred pounds will have the capacity of the human mind â€“ Hans Moravec
- 2055 â€“ $1,000 computer will match the processing power of all human brains on Earth â€“ Ray Kurzweil
Omg, we have less than twenty years for the strong AI. Don’t think so. Think also about the technology like CD burners or computer monitors. 1x, 2x, 4x, 6x, 8x, 10x, 12x… you really believe that they couldn’t go from 4x to 12x? Or from 12″ to 17″ when talking about screens? It’s about selling technology on a rates. I have a new CPU on 5GHz but i will sell it to you only after you buy 2GHz, 3GHz, 3.5GHz and 4GHz. Sorry, but it was so hard to work on it and my two hundred square meters beach house on Maldives is only half finished!
Apart from having a much better understanding of the human psychology and the functioning of the brain we need a massive parallel processing, and i am not talking about eight core processors or even a farm of multi-core processors, I’m talking about the new way of processing of the data where parallelism is the basic concept, where there could be billions of bits of information interacting with each other more like the brain is functioning. There is no way to achieve good performance on image processing and simultaneous association with other inputs (sensory information) and database (memory) with the serial algorithmic approach.
I would say, it’s good to have hope and await for the technology that would have the impact on the human society probably greater than the Internet or mobile devices had, but it’s a long road which includes many different fields of science that need to evolve before we could be able to have an really interesting conversation with our best friend, computer.