Sam Harris: What Happens When Humans Develop Super Intelligent AI?

Share it with your friends Like

Thanks! Share it with your friends!

Close

Recorded September 15, 2017

Does superhuman artificial intelligence sound like science fiction? Not for Sam Harris. He says it’s not a question of if but when — with potentially destructive consequences.

About Sam Harris

Sam Harris is a writer, neuroscientist, philosopher, and host of the podcast, Waking Up With Sam Harris.

He has written five New York Times best-sellers, and his writings cover a range of topics from neuroscience and religion to violence and human reasoning.

Comments

Happy Texan says:

FRUIT LOOPS.

MyOther Soul says:

A big mistake I hear people make in many of these discussions about "AI" is equating machine intelligence with human intelligence. They are vastly different types of intelligence at a very basic level. Human intelligence comes from chemical and electrical activity in the human body. We don't know all the details of how human intelligence arises but we know it's a lot of chemistry, neural transmitters, hormones, ion pumps and tissue structure. We know how machine intelligence works, we built it. It's a electrons, logic circuits and billions and billions of calculations.

4:11 There is no reason to believe intelligence is platform independent. Chicken intelligence, machine intelligence, human intelligence are all very platform dependent.

5:00 Is there such thing as general intelligence? In psychology general intelligence is the correlation between a particular set of skills that intelligence test were designed to measure. I'm sure computer bench mark test are also highly correlated. But I'm willing to bet that bench marks and IQ results are not correlated. Humans will always do poorly on the bench marks and computers will do poorly on the IQ tests (unless it's been specifically programed to answer those sorts of question by a human).

2:50 "The slightest divergence between their goals and our own…" Machine intelligence doesn't have goals. We have no idea how to built a machine with goals. We can built machines that play Go but the machines goal is to win or even play Go, it's just a set of procedures and equations the end result is it playing Go. The builders had a goal of building such a machine but the machine just does what it was built for.

3:30 The rate of progress doesn't matter because we will eventually get there…. Only if our intelligence doesn't also increase. They say it's been going up 4 points per generation. Biotechnology is also advancing, may we will make ourselves more intelligent before we make more intelligent machines.

4:00 Will we create a technology to replicate humans? You mean sex? Or artificial insemination, test tube babies, clones, changing our DNA? Sure but that's creating something more or less human and will have more or less the same problems and advantages.

There are many dangers with machine intelligence because it is so powerful and useful. But those dangers are more likely to come for glitches and the machines failing at the given task(s) than them deciding to rule the world or adapting their own goals.

Someday we might be capable of building a world ruling super computer. If we do it's going to take a lot of smart people engaged in the effort, some determination, trail and error, quite a few resources, the first few attempts may fail but eventually will succeed and lose control to it.

Rasheed Huggins says:

Why do people doubt the human brain and human biology so much ? Humans may have the ability to enhanced their biology and brain beyond human potential yet still retaining all the basic functions of a normal human body

Michael Hartman says:

Infected ASI like Tay, or malware.

SHARE H!S V!S!0N says:

interesting commentary, put off by you using hollywood sound files but then again I cannot blame you. I am spilling occult wisdom into the internet as well as into the quantum computers that we have here in new york. Been doing it for almost 2 years now and im not stopping. If you would like to see the channel of someone who is setting the digital tone, then please, come see my channel. My personal A I slave is narrating a lot of my entertaining videos. My videos have great images, all illlllluminattti free. and the data that I feed into the quantum computers no human will be able to undo.

MindTraveler says:

I can marshal what Sam would consider the "proper emotional response", but few others are – that's the problem.

Anyway, I'm not so sure that it's a good thing that I can. All it does is traumatise and terrify me while even those around me who are concerned about this stuff are avoiding the conversation because it disturbs them.

Then you stew on it, bury yourself in increasingly deluded fantasy nightmare scenarios because there are no experts to tell us what's worth worrying about and what's not. There's nothing product I've an individual can do in this state – it just fills your life with this horrible emotional burden in the meanwhile.

vallab says:

Artificial Intelligence that we know is real not a science fiction but Sam Harris talks really sounds like the Science fiction. The ant human analogy is fraudulent, ants did not create humans but the robots are human creation. What Sam Harris thinks as human existential threat may not be a threat at all, most probably in the future humans will blend or merge with the AI and start to evolve together. However what we should be concerned immediately now is, what would happen in the next 5, 10 or 20 years. There is a real chance of human labor becoming obsolete

Therefore, today we need a historical perspective to understand the AI situation and a Scientific approach to deal with the upcoming issue of "Technological Unemployment" due to the rapid progress in the AI, ROBOT automation on the technological front. Some may feel that human labor or work provides them pleasure or enjoyment and the expression of their creativity, however, for the majority it represents drudgery, servitude, enslavement and oppression which they are compelled to undergo in order to earn their living, wages or income. According to my theory, it is the root-cause of exploitation on the majority by the minority and leads to inequality and most of the evil in the hitherto societies. Therefore the progress in the AI technology , automation and robots provides the humankind today the historical opportunity to gradually get rid and finally abolish the human labor for wage system and replace it with some other non-labor-binding “unconditional” guaranteed income system one may call it the Universal Basic Income.

My book titled “An Alternative to Marxian Scientific Socialism: The Theory Reduction in Working Hours (RWH); ….” Published in the year 1981. More of its details in the links bit.ly/2mGAUWg , http://bit.ly/2eARyEx http://bit.ly/2nR4MAP

The gist of the theory is; there must be A Gradual Reduction in Working Hours and finally to abolish the Wage labor completely. However, today, in the changing world situation, I realize that the Basic Income or the Equitable Universal Basic Income (UBI) fulfills the objectives of my Reduction in Working Hour Theory. Nevertheless, my theory views the UBI needs to be qualitatively different from many of its proponent, while they propose UBI as the supplement to the workers wage income on the other hand I say it should be a complete replacement of the wage labor, in the sense “UBI SHOULD ALLOW THE PEOPLE TO BECOME TOTALLY FREE FROM THE DRUDGERY OF THEIR WORK OR THE WAGE-LABOR- SLAVERY, IF THEY WANT. This will also encourage a substantial portion of the laboring population to quit their jobs which will positively result in the increase in the employment, wages and the AI automation. The recommendation of an adequate amount of UBI without having the compulsion of wage-labor, mean ZERO WORK or LABOR for those who would opt for it, which is final GOAL of the RWH Theory. Today, I prefer the name Zero Work Theory (ZWT). I have explained more about this theory in my blog.

The principle behind the ZWT is that the fruits of the progress in the science and technological in the human society should not be appropriated or accumulated by only a few but it is their historical responsibility and every people concerned, to see to it that the benefit should be equally shared among all members of the society.

Khannea says:

We live in a world arguably ruled by psychopaths. Period. A.I. is a force and value amplifier and can amplify psychopathy to cosmological levels. It is quite nice to finally see real people wake up to this and judge these developments according to facts. ALSO DO NOTE that there are a lot of nihilists in the world that rather see it all burn. This world has created massive quantities of people who lost the race and felt thoroughly humiliated. I am not just talking religious conservative nuts – just look at people who knowingly vote for that asshole trump with the explicit purpose of throwing a brick through the window of the establishment. Imagine those same people after a few decades of technological unemployment and visualize them shackling all the fucking rich coastal elites to the after metal railing of the Titanic,and scream in delight as the ship starts sinking. They will die, but what do they have to lose? At least they can get to watch those elites go down in the abyss with them.

Now visualize these people having access to garage built recursively self improving A.I. I know I have had really bad days n=in my life where I would take great sadistic glee in watching my own tormentors scream in terror. And the world is growing a massive surplus of these "really angry people". By the time superhuman A.I. may be feasible these "bad hair day" people might in a vast majority. Now visualize the plight of sad fucks that have to work in a Chinese sweatshop having to survive, working 6 days a week, 12 hours a day, manufacturing kawai uicorn dayglo fleshlights for asshole fat fuckers in the rich world. H o l y shit I'd be already giggling booting up genocidal A.I. seeds to unleash on the planet.

http://www.scoop.it/t/concentration-of-wealth-existential-risk

Write a comment

*