953 replies to this topic

[Major Legend]

Sorry I made a big mistake - I am human after all, and not a very bright one at that. I meant ANI . I always mistake ASI as Artificial Specialised Intelligence, but actually it means Artificial Superior Intelligence. What I meant was ANI ( Artificial Narrow Intelligence) will long replace most human jobs before AGI even comes to reality.

 

What I mean is various human functions such as motor coordination and vision will long be surpassed by ANI well before AGI is even developed. When AGI is finally developed it will be ASI once it taps into the collective data and abilities of ANI capabilities.

 

So please reread my post! It's the wrong way around! : D

 

Also one biggie about why AlphaGo was huge was that playing Go requires abstract intelligence, the kind of thing which makes you think oh that's a nice piece of art, chess is very logicial, thats why nobody predicted an AI could beat a human in Go that quickly.

^ The issue is humans are vastly over estimating their own intelligence and competency in things, whilst we do exhibit superior intelligence, most of our functions can be broken down and performed with a computer/robot.

 

Deep learning allow computers with neural networks to learn vastly faster than humans at almost any task, whether that's throwing a basket ball or landing a jet plane. The missing piece is computer vision as as we have seen from Microsoft and Stanford, that's already becoming possible using deep learning. It isn't hard to imagine a computer that would become better at identifying objects and things rather than people.

 

Ironically the largest limitation of ANI is the inability to perform generalised tasks such as assembling a product like an ikea together without many specialised machines, but this is also where the manufacturing process is the cheapest which is why it gets outsourced to other countries, and who knows with 3D printing maybe manufacturing isn't even such a big challenge for the society of the future.

 

Modern pilots are nothing but backup systems, and most plane crashes are caused by a pilot negligence rather than a computer problem. There is just no competition, computers don't even need sleep, their abilities don't decrease when they are tired. Whilst it's great for businesses to stay "moral" by employing people it isn't practical under global hyper-competition, as we have seen companies are forced to either use cheap labour or just automate the whole thing entirely.

 

edit: Whoops I deleted my own post. I thought Stop! might mean being able to edit my old post, now the post is gone   can a moderator help me restore the old post with ASI being corrected to ANI.

 

 

Edited by Major Legend, 18 April 2016 - 01:14 PM.

[mag1]

The MOOC revolution is underway!

This will profoundly change the life experience of almost student on this planet.

The implications for global employment are also substantial.

 

The extreme power to choose how and under conditions children are educated that until now has largely been

the preserve of those with wealth, will soon be a birth right for all. The appalling sociological conditions present

in many government funded schools and which have directly caused so many terrible tragedies will soon be a 

thing of the past.It will be fascinating to watch this new era of choice through online education unfold. 

 

https://faculty.ist....0Kbachelors.pdf

http://edf.stanford....e-seen-official

Edited by mag1, 21 April 2016 - 01:57 AM.

[mag1]

Breaking up the existing power grid with a decentralized solar economy would strand a huge quantity of capital.

 

The price of electricity could be heading towards near 0!

Once residential solar is deployed, with a marginal cost of 0, and probably a huge oversupply of power

during daylight hours.. well free electricity! 

 

http://en.apdnews.co...hua/391330.html

[niner]

Breaking up the existing power grid with a decentralized solar economy would strand a huge quantity of capital.

 

The price of electricity could be heading towards near 0!

Once residential solar is deployed, with a marginal cost of 0, and probably a huge oversupply of power

during daylight hours.. well free electricity! 

 

http://en.apdnews.co...hua/391330.html

 

Um, no.  That's like saying that a new car is free, as long as you ignore the purchase price.   The cost of electricity has to cover the amortized cost of the cells and other equipment needed to produce it, along with the cost of maintenance, repair, and replacement.  You also have to provide for storage so that you have power when it's dark.   The linked article is interesting in that they've made a cell using materials that are not highly toxic, but the efficiency isn't high enough to commercialize yet.  There's no indication that the cost would be lower per watt than current commercial cells.
 

[mag1]

The dismal science just got dimaller.

Agriculture used to be the textbook example of a competitive market in which there was no price floor.

 

Welcome a new candidate to the textbooks: electricity.

Consider an electricity market in which everyone has the ability to put spare power onto the grid.

With fairly inelastic demand and near perfectly elastic supply, the rush to zero kwh seems almost a certainty.

 

In the electricity market the price of the solar cells is a sunk cost.

Sunk costs are ignored by rational businesses.

 

What should the price be?

With possibly tens of millions of producers located in what might become a globally interconnected grid, a startlingly modest

price does not seem unlikely. In fact, this might be the perfect opportunity for guerilla green ecoecons to step up: drive the cost of electricity to near zero with solar cells. With solar cells there is at least some return (excluding the sunk costs) even near 0 cents per kwh. In fact, such economic activism would have enormous leverage. As soon as you go slightly above the quantity demanded in the electricity market the price could rapidly collapse. Natural gas, nuclear and coal ... wouldn't make economic sense nearly instantly.

 

Oh and electricity has an interesting economic property.. use it or lose.

It is the ultimate no fold bluff.

You will not accept my offer of .01 per kwh?

OK, poof there's a big donut for you.

Want to play round 2? 

 

Oil markets have shown time and time again what can happen when a near essential commodity has even slightly limited

supply. Electricity markets would show the reverse effect. What would happen in the electricity markets if people wanted

1000 gwh and there were 1001 gwh available? Consider also that the marginal cost of solar power is near zero.

 

Further reports on the above discovery talk of it being prepared to launch commercially over the next year/s time frame.

It is expected to have mass uptake in residential markets etc.

 

 

Edited by mag1, 30 April 2016 - 08:00 PM.

[niner]

You will not accept my offer of .01 per kwh?

OK, poof there's a big donut for you.

Want to play round 2?

 

No one is going to spend $25K to build a solar installation big enough for a typical American home if there's a prospect of nearly free electricity on the horizon.  You're counting on an excessive build-out to the point of oversupply, along with a sophisticated (costly) grid in order to create nearly free electricity, but people are just not going to build it without up-front prices falling radically.   The sun only shines during the day, so electricity that isn't needed in daylight will be stored for night use.  excess electricity can always be used; the production of aluminum or many other industrial processes can take advantage of an oversupply.

[mag1]

I really do not see how we are not going to arrive at a 0 kwh price point.

Even now all sorts of Power Points, pie charts, and net present value calculations are based on this fairly dubious assumption of future

electricity prices out 5-10 years. Some of these calculations even extend out to making assumptions about what

electricity prices might be in 20-25 years. Should these calculations not consider that solar cell panel costs have been in free fall

for many many years? Price parity has already been reached in some markets. At this time and given the ongoing technological advances, I would not be comfortable projecting electricity prices more than 2 years forward.

 

Electricity monopolies will want to maintain their prices, all the while there will be a buildup in the generating capacity.

Everyone is happy until the oversupply can no longer be ignored.

 

And then it is back to 0.1 per kwh or donut. 

 

How many more bubbles need burst? 

Do we never learn from our past?

 

There are even now a fairly large number of people who are happily putting down their $25K for the solar install.

Government seems to be quite happy to throw tax credits around.

 

A year or two ago we lost power for a few days during the middle of a very cold winter.

It would certainly be in our rational self interest to pay $1000 for a solar emergency power back up,

even without regard to what power prices might be. If this became a standard feature on new homes,

then the collapse of energy prices would become certain.

 

The world's largest solar plant financed with government dollars is nearly ready to go online in Buffalo.

Considering the nearly continuous technological change that has happened with solar cells over the last century,

it takes quite a bit of courage to put new money into solar. Might be why so much of this investment is non-private.

 

https://www.technolo...ys-gigafactory/

Edited by mag1, 30 April 2016 - 10:35 PM.

[niner]

Mag, that new gigafactory is projecting costs (to the company) of $2.50 a watt.  The cost to the customer will be more, and those watts are during maximum insolation, at the brightest part of the day.  You'd need way more than $1000 to get serious power, (consider the cost of installation, inverter, and switching, and that's not even considering storage), although a couple hundred watts is certainly handy.  The "collapse of energy prices" that you're predicting is not even remotely close.  It would take more than a slightly cheaper or slightly more efficient solar cell to get there.  We would need a major technological breakthrough that hasn't happened yet.

[marcobjj]

Moore's Law is predicted to flatten out in 10 years. Due to heating issues, silicon chips cannot surpass the protein based human brain in data processing. I do believe in AI taking over in the distant future, but won't  be a reality for the next 100-200 years. Could take longer. Like a pen and a piece of paper was the state of the art data storage for millenia, we could be stuck with these primitive silicon chips for a while.

 

 

 

Edited by marcobjj, 01 May 2016 - 04:17 AM.

[marcobjj]

At the peak in 1979 America had almost 20 million manufacturing jobs. Over the last 15 years America has lost 5 million factory jobs and now has 12 million (and falling).

Those lost jobs were probably replaced with lower paying service jobs.

 

Am I the only person who was unaware of this?   

 

 

Robots didn't take those jobs, the Chinese did. US Manufacturing was outsourced when China entered WTO in 2001. And hence why D Trump wants to bring it back. Coincidentally, US economy hasn't experienced real growth since.

[niner]

 I do believe in AI taking over in the distant future, but won't  be a reality for the next 100-200 years. Could take longer.

 

How can you possibly predict what will happen in 100-200 years?   Could you have predicted the Internet, or SENS, or CRISPR, in 1916?  1816?

[mag1]

Solar energy prices seem to just keep falling and falling!

The price point quoted below is 50 cents per watt for the modules.

The wiki article is talking about prices in 2014.

The numbers are moving around so fast it is hard to know where the price right now might be.

 

Of course that does not even include what might now occur due to the breakthrough that I posted above. 

 

 

http://pv.energytren...ricequotes.html

https://en.wikipedia...ovoltaic_system

http://michaelblueja...city/solar.html

 

The idea for the $1000 home unit was more for emergency backup.

One of the most frustrating things during the blackout a few winters ago was that all we needed was an electric ignition to get our natural gas water heater up and running. This could have been a great benefit to us. Perhaps we could have set up some piping to heat our house off the water heater.

 

If I were sitting in the big chair right now, I think that I might mandate some sort of emergency solar power for all new home construction. It does not seem unreasonable in this day and age to require homes have some sort of independent energy generating capacity. This forthcoming new technology in solar power using non-toxic cheap components might allow us to reach this goal.  

Edited by mag1, 01 May 2016 - 04:50 AM.

[marcobjj]

"How can you possibly predict what will happen in 100-200 years? "

 

It's just a guess in the end. But only quantum computing can surpass human intelligence, and quantum computing doesn't look like it's within reach until at least the end of this century.

Edited by marcobjj, 01 May 2016 - 04:51 AM.

[mag1]

I think the "Exponential curves are linear in log scale" makes a whole bunch of sense on this question.

It seems that most experts are tricked on this one.

 

The price per base pair example is an extreme example of this.

Experts were thinking it could take centuries to reach the mass genomics era.

Yet here we are 15 years later and the $100 genome no longer seems impossible.

 

I would agree though about computer chips.

I just recently bought a new computer and I was quite shocked to realize that this new computer was no better and not much cheaper

than a computer that I bought 3 years ago. I have started to wonder whether the idea of converting the universe into one enormous

computer might be somewhat fanciful. Trees do not grow through the clouds.. computer chips might have already peaked out.

 

It is not so much that we couldn't reach the stars, though people might just give up and think we did pretty well.

Without an economic rationale we might never reach the Singularity.

[marcobjj]

 

I would agree though about computer chips.

I just recently bought a new computer and I was quite shocked to realize that this new computer was no better and not much cheaper

than a computer that I bought 3 years ago. I have started to wonder whether the idea of converting the universe into one enormous

computer might be somewhat fanciful. Trees do not grow through the clouds.. computer chips might have already peaked out.

 

 

yep. You can also witness this slowdown in video games. 8 years elapsed between the x360 release 2005 and the xbone in 2013. The longest interval between two generations. The the graphical improvements are pretty marginal when compared to the jump between previous gens. We are close to a computing plateau.

[Major Legend]

 

 

I would agree though about computer chips.

I just recently bought a new computer and I was quite shocked to realize that this new computer was no better and not much cheaper

than a computer that I bought 3 years ago. I have started to wonder whether the idea of converting the universe into one enormous

computer might be somewhat fanciful. Trees do not grow through the clouds.. computer chips might have already peaked out.

 

 

yep. You can also witness this slowdown in video games. 8 years elapsed between the x360 release 2005 and the xbone in 2013. The longest interval between two generations. The the graphical improvements are pretty marginal when compared to the jump between previous gens. We are close to a computing plateau.

 

 

Yes moore's law will flatten, but we have been finding far more creative ways of using processing power. For example cloud computing has allowed access to far greater processing abilities on small devices like smartphones than otherwise possible. For example voice assistants have become a lot more powerful from 3 years ago. Neural networks and deep learning is likely to leverage the physical limitation, much like the way our brain does. Considering our brain is still far ahead in terms of ability and storage compared to current day computers - I think the real physical limitation is still pretty far away. 

 

The console model is not really relevant because the problem is with developers being unable to utilise all the processing power to design games, the better looking the game the more polygons and work needs to go into the games, however if you look at procedural generation in games like No Mans Sky, we are reaching an incredible level of world creation, because it's really about what you do with the processing power. If you look at the current games and software out there the vast majority barely even scratch the performances of the highest performing computers. Also the polygons have been increasingly pretty insanely it's just the way polygon works 2500 to 5000 is going to look a lot more insane than 5000 to 10000, and again the limitation does come down to the fact that all the graphics need to be hand designed to look photo realistic.

 

^ Also there has simply been less interes in consoles because of the intense diversification of platforms and the inability for games to stay "exclusive". In the past a certain game would only run on one console giving a huge incentive to the winner of a console war, now people are split across a bunch of things like PC, smartphone, tablet and console meaning there is far less incentive to have close back to back complete generational revamp. If you look at the current Playstation 4 the technology is about the same as a cheap low budget PC, also case and point - the PS4 still has loads of unreleased games, the latest Final Fantasy for example, hasn't even come out yet. Another game is Detroit: Become Human (Pic attached) - which is the only game that peaks the PS4's abilities.

 

if you used methods like higher textures and smoother modeling you will just arrive at a better "remastered" object and not really one that looks more photorealistic. You can test this by finding some old games like Final Fantasy X and see it being remastered on new consoles with better textures, it still looks like a old generation game.

 

This is different from when you jumped from the Super Nintendo to the Playstation 1 and all of the sudden developers can do things they wanted to do before but couldn't, now it's more of an issue that you need X amount of millions of dollars and an X amount of years to make something, and then you have to be sure to be able to recoup that expenditure, hence most people just aim lower. At the end of the day every line of code has to be written and every graphical asset needs to be drawn somewhere, until a computer can do that you will not really see this change. 

 

So I think you are looking at a "market limitation" rather than a physical limitation - cheap PCs almost do everything people want, the problem with high end PCs is there simply isn't really a need to justify spending money on it, so hardware as a commodity has generally been in decline leading to less development, if you look at other technologies such as camera, visual recognition, sensor miniaturisation and so on the progress has been pretty incredible because people want better and better phones, but even that is starting to hit a market snag.

 

In terms of physical limitations imposed by quantum tunneling there are already designs for 3d stacks of chips and other ideas to bypass this problem, sure the increase won't be exponential, but it's still going to keep going and going, the bottleneck seems to be human designers rather a computational power. For example developers have found it difficult to program for multi-core processors, because frankly its a pain in the ass to do so, however software will keep progressing exponentially meaning deep learning programs will be able to create programs that would have taken years or decades in the past, we are already seeing this trend with an AI beating a world class Go player, automated driving systems and frankly a whole load of scary things that indicate processing power superior to human intelligence at least on a narrow field.

 

 

Attached Files

•  Detroit-Become-Human-2.jpg   49.21KB   0 downloads

Edited by Major Legend, 01 May 2016 - 08:19 AM.

[PWAIN]

Exponential curves are actually a series of 'S' curves in the real world. Each 'S' curve is a different paradigm. In the example of computing, you start with the manual paradigm, abacus followed by manual calculator, then valves, discreet transistors, discreet logic chips and finally lsi chips. What comes next is not yet obvious but there are a number of candidates. We might have a slow down until this emerges but then we are likely to see huge acceleration.

[marcobjj]

 

 

 

 

 

In terms of physical limitations imposed by quantum tunneling there are already designs for 3d stacks of chips and other ideas to bypass this problem, sure the increase won't be exponential, but it's still going to keep going and going, the bottleneck seems to be human designers rather a computational power. For example developers have found it difficult to program for multi-core processors, because frankly its a pain in the ass to do so, however software will keep progressing exponentially meaning deep learning programs will be able to create programs that would have taken years or decades in the past, we are already seeing this trend with an AI beating a world class Go player, automated driving systems and frankly a whole load of scary things that indicate processing power superior to human intelligence at least on a narrow field.

 

 

3d stacks of chips are still integrated circuits however, they are not quantum computing. Listen, the point was, this generation of silicon computers won't surpass human intelligence, or come close, no matter how much we tweak it. 

 

Google's computer defeating the top Go Player is cool, but it's not a revolution over the Deep Blue computer defeating Kasparov in a Chess match in 20 years ago. It's still just a board game, discrete X,Y coordinates in 2D space. That is still not close to replacing the most menial of jobs. What your hispanic housekeeper does is at least a million times more complex. 

[PWAIN]

Marcobjj, if you think go and chess are in any way similar then you have no idea. When chess was won, many experts commented that it would take at least 100 years for go to be won. It took 19 years! The reason it only took 19 years is because we used neural networks, these NNs are quite a new type and allow problem solving without going through all the possible moves, ie. They are starting to 'think'. The chess work was more about searching known moves and going through as many moves as possible.
No one knows what will replace today's silicon chips but it is quite possible they will be able to squeeze a lot more out of it. Another possibility is silicon being used as the basis of a new technology. Quantum computers don't work in the way you seem to think they do. They are not one of the better candidates to replace silicon because they cannot be used for general purpose computing.

[niner]

But only quantum computing can surpass human intelligence, and quantum computing doesn't look like it's within reach until at least the end of this century.

 

"at least the end of this century"?  Seriously?  Small quantum computers exist right now.  However, I don't see any reason why you would need quantum computers to surpass human intelligence.  The human brain does it with a large number of very slow components.  We already have voice recognition that's better than humans on speech fragments.

[marcobjj]

the most complex operation done by a quantum processor was 3 x 5 = 15 (with lower than 50% accuracy). That was in 2012. A couple of years ago went up to 56,153 using two factor.  

 

In terms of usability, quantum computing is behind the Blaise Pascal Arithmetic machine from 1642. Yeah, it could take a while alright.

 

 

 

 

[niner]

In terms of usability, quantum computing is behind the Blaise Pascal Arithmetic machine from 1642. Yeah, it could take a while alright.

   

You sure about that?

[marcobjj]

Yes, but what the video alludes to is the development of a universal quantum computer that can solve any computing problem. D'wave solves linear equations. it also requires a two story building just to cool off its chip. So it doesnt circumvent the heating problems that have halted the progress of silicon based comventional chips.

https://m.youtube.co...h?v=m3TOWanwuO8

Edited by marcobjj, 01 May 2016 - 10:24 PM.

[marcobjj]

^"conventional"

[niner]

Yes, but what the video alludes to is the development of a universal quantum computer that can solve any computing problem. D'wave solves linear equations. it also requires a two story building just to cool off its chip. So it doesnt circumvent the heating problems that have halted the progress of silicon based comventional chips.

https://m.youtube.co...h?v=m3TOWanwuO8

 

Dwave does a lot more than just linear equations.  Who the hell cares how large the refrigerator is?  I think they've taken care of the heating problem adequately, and small increases in the number of qbits exponentially increase the machines capability, so they don't need billions of units on a chip like a conventional computer does.   Which, I might add, is completely beside the point.  You said it would take a century to develop this thing that exists right now.  Just check your facts before you post, and save us all a lot of trouble.

[marcobjj]

D-Wave's quantum computer, rather, only solves optimisation problems, that is ones that can be expressed in a linear equation with lots of variables each with its own weight (the number that is multiplied times each variable). Normally, such linear equations are very difficult to solve for a conventional 'universal' computer, taking lots of iterations to find the optimal set of values for the variables. However, with D-Wave's application-specific quantum computer, such problems can be solved in a single cycle.

"We believe that starting with an application-specific quantum processor is the right way to go, as a stepping stone to the Holy Grail, a universal quantum computer," Hilton noted. "And that's what D-Wave does," dealing with optimisation problems using qubits."

http://m.eetindia.co...TM#.VyfCE9RHarU

Yes, but what the video alludes to is the development of a universal quantum computer that can solve any computing problem. D'wave solves linear equations. it also requires a two story building just to cool off its chip. So it doesnt circumvent the heating problems that have halted the progress of silicon based comventional chips.

https://m.youtube.co...h?v=m3TOWanwuO8

Dwave does a lot more than just linear equations. Who the hell cares how large the refrigerator is? I think they've taken care of the heating problem adequately, and small increases in the number of qbits exponentially increase the machines capability, so they don't need billions of units on a chip like a conventional computer does. Which, I might add, is completely beside the point. You said it would take a century to develop this thing that exists right now. Just check your facts before you post, and save us all a lot of trouble.

Not in within the context of this topic, which is specifically about robots replacing human labor. In wich case size matters, just as type writers didnt become obsolete the day the military unveiled the EINIAC in the 50s, it took several decades to make it portable.

Edited by marcobjj, 02 May 2016 - 09:18 PM.

[niner]

Not in within the context of this topic, which is specifically about robots replacing human labor. In wich case size matters, just as type writers didnt become obsolete the day the military unveiled the EINIAC in the 50s, it took several decades to make it portable.

 

Quantum computers are in no way needed to create robots that can displace human workers.  Self-driving cars are about to displace lots of workers, using nothing but plain old silicon.   You brought up quantum computers with your claim that only quantum computers could exceed human intelligence, which is completely unsupported.

[mag1]

Hmm, let me bud in here a bit.

 

I actually have found this discussion about the current state of quantum computing quite edifying.

It was amazing how helpful it was in quantum mechanics courses when someone would put up their hand and ask

for a clarification when the professor said "So, obviously..." It wasn't obvious to anyone in the room except the prof.

 

I found the reference to DWave very helpful.

It is great to have an update from them every year or so.

I mean it would be best to keep an eye on those who might have built a computer 100 million faster than what we have now.

We could be pushed into the Singularity before there is even a time for a good old fashioned protest against the man, the power, the machine? ... .

 

It also got me thinking about a possible scifi plot.

The comment about a robot needing it's own attached brain seemed off to me. Why would a robot need it's own brain?

Why not have a disembrained robot? Robot here <-- brain there -->.

 

Maybe what we could have is the quantum brains being located in one place-- say, Canada.

Canada is cool.

Canada is coooler than say Nevada.

And Canada is cooler than California.

Why not have the entire global QC brains in Canada? Comparative advantage and all that.

 

Sounds like an interesting premise for a book.

Could things go wrong? Sure.

Maybe Y2K for real.

Well, that is why books have plots.

 

 

 

 

 

 

 

 

 

Edited by mag1, 03 May 2016 - 02:39 AM.

[marcobjj]

nope, take a look at the title again: "automation and AI will take most human jobs". Claiming that will take some is nothing new and has already happened in the past.

It is my personal belief that only quantum computers can surpass human intelligence. That the progress of moore's law, and thus AI, will flatten out in the silicon generation is what most think.

[Major Legend]

 

 

 

 

 

 

In terms of physical limitations imposed by quantum tunneling there are already designs for 3d stacks of chips and other ideas to bypass this problem, sure the increase won't be exponential, but it's still going to keep going and going, the bottleneck seems to be human designers rather a computational power. For example developers have found it difficult to program for multi-core processors, because frankly its a pain in the ass to do so, however software will keep progressing exponentially meaning deep learning programs will be able to create programs that would have taken years or decades in the past, we are already seeing this trend with an AI beating a world class Go player, automated driving systems and frankly a whole load of scary things that indicate processing power superior to human intelligence at least on a narrow field.

 

 

3d stacks of chips are still integrated circuits however, they are not quantum computing. Listen, the point was, this generation of silicon computers won't surpass human intelligence, or come close, no matter how much we tweak it. 

 

Google's computer defeating the top Go Player is cool, but it's not a revolution over the Deep Blue computer defeating Kasparov in a Chess match in 20 years ago. It's still just a board game, discrete X,Y coordinates in 2D space. That is still not close to replacing the most menial of jobs. What your hispanic housekeeper does is at least a million times more complex. 

 

 

I agree we won't replace a hispanic housekeeper anytime soon, by the time we have such technology, computers would effectively become sentient or super sentient, so no - but what does a housekeeper do? A robot navigating a house isn't unimaginable with current level technologies, the missing key is visual recognition which is developing rapidly now. Cleaning the floor, a Roomba can already do that and they are getting better with each generation.

 

Some other things like folding clothes would be more difficult, but the point is ALOT of jobs can be replaced, and usually it is the low paid stuff like being a housekeeper or cleaning the toilets that machines have difficulty doing.

So I think the question isn't so much for me whether a computer can surpass a human in absolute intelligence, it is whether computers/machines can surpass humans in most narrowly focused jobs, and I think the answer there is yes. At least the displacement would be large enough to have a big impact in society.

 

Even IF a computer surpassed a human it wouldn't act like one unless you can put in 18-20 years of human life programming in it. I think the line is blurry here, there isn't like a specific point to look for.

 

Maybe designing a game or a website is a huge use of human exclusive intelligence, maybe accounting and law does, but maybe/probably not. For example a lot of transport can be automated that means displacing drivers, factories can continue the trend onwards full automation. You have software right now that can primitively edit videos automatically, when you add in AI, I don't see why silicon based computers won't eventually be able to replace video editors. 

^ Think of all the creative and professional jobs like designers,pharmacists, directors, stock market analyst, programmers, pilots, surgeons - these people think they are not at risk, but is what they do really really that complicated?  They follow instructions, learn, and use abstract skills to do a task (a lot of times poorly) - if you made a computer purely designed to do one job, everything says to me that a Human just won't be able to beat it.

^ I remember reading an article about how drivers can't be replaced because of the combination of cognitive skills required to drive, but then Google and Tesla did it. I think we often take pride in our work and it's difficult for us to admit that a lot of human jobs just don't take as much brains as we like to think.

Of course there will be people who can't be replaced, people like CEOs, administrators, land owners, engineers of these systems and so on, but the problem is you only need to replace 20% of the population to effectively collapse the current system we have.

So back to the housekeeper that gets paid below minimum wage - sure computers can't do that particular job, but would you want to do it? 

Edited by Major Legend, 03 May 2016 - 03:48 AM.