Yes, years. Libratus uses 16 Terabytes of memory for just a 2 player game. The size of the game tree increases by at least a factor of 1000 when moving up to just 3 players.
Considering this AI is already dealing with unknown information, I doubt the size of the "game tree" increases by the factor you cited as you add more players.
There is no sign of a death spiral in these layoffs. The article cites 1600 job openings Microsoft currently has posted on LinkedIn alone, so they are still growing. And each of these employees has 60 days to find new internal positions at Microsoft, so if their skills are still relevant they don't necessarily even need to leave Microsoft.
This is simply sign of a transitioning company that knows it cannot be successful simply sticking to the status quo.
I believe that's a bluff and I'm prepared to call it. If they were going to do that, it'd already have been done.
What are you talking about? Offshoring of IT work already does happen, and will continue to happen. If they haven't done it already, it is because of the cost of doing it (cost in either money, effort, and/or quality). The American workforce may still benefit from limiting H1B visas, but it will almost certainly have an effect of increasing the amount of offshoring. The hope is we offshore less jobs than we save, but it's far from a guarantee.
Considering it's nearly impossible to find good software developers these days, I find it hard to believe things would be better with even less software developers in our local economy.
Well, Obama was at war for all eight years of his presidency. Beating all other presidents. Hard to argue that he was better than Bush with that record.
How is this insightful? Obama was cleaning up the mess of Bush's wars and Obama somehow has a worse record on warmongering than the guy who started them?
Put another way, 4222 US soldiers died in Iraq under Bush's watch, and 292 died under Obama's watch (that figure includes people who died in January 2009 before Obama took office). Claiming Obama had a worst war record than Bush is asinine.
Loans do not count. Of course you can get loans... that cannot be discharged in bankruptcy.
Why don't loans count? My ability to take out about $100k in loans with no collateral and horrible credit is directly responsible for my income rising from $40k to $160k in less than six years. I significantly turned my life around starting in 2009 (at age 28) primarily because of student loan availability regardless of credit rating or past academic history (which was also horrible for me at the time), which is only possible because they cannot be discharged in bankruptcy.
Make up your mind. Which is it? They are paying more to white males or discriminating against white males by preferring Asian workers?
Oracle has not been complying with requests for information, so the Labor Department is suing them so they can get this information during discovery. They are probably suing them for multiple offenses so they can get a more broad set of information during this process. At least that is my guess based on the information in the article.
"The brain is just a computer" is hand-wavy and objectively incorrect.
No it is not. In the context of the rest of the paragraph it is obvious he doesn't mean a digital computer with an x86 instruction set. He was referring to the brain as a machine which accepts inputs, processes the inputs based on it's current physical state, and produces various outputs to initiate action throughout the body. He mentioned Substance Dualism at the beginning of the paragraph the line you are quoting is found in order to clear up any confusion about what he meant by calling the brain a computer.
I will grant you he is using a definition of computer not used in the common English lexicon, but since we are talking about potentially future human-level AI it's likely the definition of computer in common usage could change by then to include biological computers, not just silicon.
I read through the paper, but couldn't find any description of what the brain does which couldn't be considered information processing. It may not be digital processing, and it may not resemble how computers process instructions and retrieve data, but even if the physical architecture is different the brain still seems to be processing information.
He uses an example of a dollar bill, and how a person cannot recall every detail of a dollar bill from memory if asked to draw one. And that is somehow proof that the brain does not store data about the dollar bill? The person was still able to draw some details about the dollar bill, such as a person in the center and the numbers in the corners, so that data was stored somewhere. The author also makes a silly distinction between "storing data" and "changing the brain", as if the way the brain is changed isn't how it stores the data.
But neither the song nor the poem has been ‘stored’ in it. The brain has simply changed in an orderly way that now allows us to sing the song or recite the poem under certain conditions.
Sounds a lot like the brain stored the song or poem somewhere in a way it could be retrieved later, under certain conditions. Just because the brain stores information in a less precise way as a computer doesn't mean it isn't storing anything. The rest of the article continues to make similarly odd claims without backing them up. The researcher takes some very valid arguments about how many researchers rely too heavily on computer / human brain metaphors, but then he makes a lot of wild statements himself without backing them up either.
And all of those have been branches of the AI field. Since the field of AI was created, arguably during the 1956 Dartmouth workshop, it has included topics such as expert systems and statistical methods being used to make decisions based on both simple and complex input. Whether you want to accept it or not, even those Pacman ghosts' behavior can be accurately referred to as AI.
If you want a term which only refers to human level intelligence, perhaps you should use either Artificial Consciousness, Machine Consciousness, or Synthetic Consciousness. Those probably match your personal definition of AI better than the broad definition used by the scientific community.
It is bad because this type of research could lead to less availability of movies on streaming services. If the studios have hard proof that Netflix is costing them money, why would they allow their movies to be shown on Netflix? Either we would see far less movies available, or the prices would go up.
Funny how Benioff mentions his Einstein feature when mentioning how much deep AI is already being used without people noticing. In this case, it would be very hard to notice since Einstein isn't even a live feature of Salesforce yet. Saying the technology is already pervasive, and then using an example that is still around the corner, is very disingenuous.
But then again, this was just Slashvertisement anyway.
That is just sensationalized reporting. The eight people in that report own $426.2 billion in assets, meaning those bottom 3.6 billion people have an average of $118 in total assets. That means the median US retiree has more wealth than 1500 "average" people in the bottom 3.6 billion poorest humans. I guess it's kind of pathetic that the average US retiree thinks they are more important than 1500 people by amassing such wealth, or some other such nonsense.
Heck, you only need a net worth of $10 to have more than the bottom 100+ million Americans combined. That is why these types of statistics are meaningless, and are only useful for grabbing headlines.
Right. Ignoring the fact that 99% of stock is owned by the 1%, so that the collective ownership of stock by the other 99% of the population amounts to approximately nothing. It's not ownership if you have no voice in its dispensation and the only thing you can do with it is sell it to someone else.
I see no technology in the near term future that I think has any reasonable probability of causing mass unemployment greater than we've seen in previous generations and in previous technological eras.
Natural language processing, self-driving vehicles, and improved virtual assistants for starters. I'm not saying they are certain to cause mass unemployment, but they certainly have a reasonable probability of doing that. Job displacement caused by software create job displacement at a much faster rate than those caused by robotics, because they are deployed at a much faster rate.
So maybe we should stop calling Machine Learning 'AI'. Shall we, pretty please?
Why would we? Should we stop calling a Corvette a car? Machine learning is a subset of AI. Ever since the AI field began it involved everything from the most rudimentary rules engines to the promise of general intelligence. Only people who get most of their understanding of AI from movies think it only means Skynet.
Yes, but there are also millions of former manufacturing and other low-skill workers who cannot find work in the new economy. No one is saying everyone will be out of a job. And it doesn't even take a majority of people out of work for there to be a problem. All it takes is a small disruption to cause massive problems.
The first industrial revolution was hugely beneficial overall to workers and company owners.
Yes, eventually. But while it is easy to look at the period from about 1760 to 1840 as a small blip in history, that was eighty years where large groups of people were significantly negatively affected by changes in employment. It's also easy to look at the century where farming went from a majority of the workforce to only a few percent of us as an easy transition, without looking at how rural areas are still dealing with the loss of income and jobs today. Manufacturing came in for about half a century to help the transition, but there isn't another savior in the horizon (at least in the short term).
We already know new jobs are almost never created fast enough to help displace workers.
You can put that idea to bed by looking at employment rates.
What are you talking about? Workforce participation by working age adults is dropping fast. We are at levels not seen since the 1970's, when women participation was half what it is now. The numbers are clear an unambigious, and they point to a large portion of our country that is being displaced by technology. It is already happening. People are only worried that AI will make the problem worse, not create a new problem.
if automation lets you create twice as much value in the same time, then there's a strong argument to be made that you should get paid twice as much rather than the executives and shareholders pocketing the difference.
There isn't even a weak argument for that let alone a strong one. If investments into automation paid for by the company are causing the increase in value each employee can produce, why do the employees deserve increased pay? If the company had paid for two human assistants for each existing employee, and their quality of work improved because of it, should the employees deserve more pay then as well? Because that isn't just an analogy, it is literally the same thing as automation improving productivity.
The truth is, people don't know what they're talking about when it comes to AI. AI will be a thing in the late 22th century, perhaps. As of now, Artificial Intelligence is just a buzz word to entertain the clueless at Corporate conventions.
No one is talking about a Skynet-level general AI when they are talking about the AI which will take someone's jobs in the next few decades. They are talking about improved voice recognition, image classification, machine learning algorithms, etc. These are the technologies threatening jobs in the short term. We don't need AI robots with consciousness for workers to be displaced.
You could say that about any technology. Most of the hand wringing over AI taking everyone's jobs is the same sort of paranoid response we've had to every technology improvement. We've seen this play before. Back in the 1970s everyone was convinced industrial robots were going to take their jobs tomorrow. Robots did become an important tool but it took decades and most of the displaced workers found new employment in comparatively short order. And plenty of people are still employed on the assembly lines right next to those robots they worried about.
the threat is it will increase productivity rapidly enough to replace 20%+ of workers quickly enough that new jobs won't be created fast enough to offset the losses.
While I think your numbers are suspect, this is the only rational argument worthy of concern. There is literally an unlimited amount of work to be done but it takes some amount of time for people to adjust to new economic realities. I think that people are vastly overestimating the risks involved here but fast increases in productivity make for short term dislocations in the work force. Some people have a hard time keeping up.
In the first industrial revolution it took generations for workers to recover from crippling job losses due to new machinery. These are not short term dislocations. There may be a nearly unlimited amount of work to be done, but not an unlimited amount of work which produces enough economic value to support someone financially. I would like someone to clean my home office, for instance, but I'm not paying $15/hr for it.
No one is making predictions without strong historical and current indicators showing there is already a problem. We already know new jobs are almost never created fast enough to help displace workers. We have just been lucky enough that only small sectors of the economy has been affected at once in the past, so we can easy ignore those who suffered.
The vast majority of human history, at least 99%, was of a world where nearly everyone was in poverty by modern standards. Thinking that our current shared prosperity is somehow the status quo which can be sustained with minimal effort is ridiculous.
It shows such a lack of understanding of the problem when he says the industry should focusing on saving people time instead of replacing people. Saving workers time so they can be more efficient is what allows companies to cut staff. Saving time and working more efficiently is the whole reason AI threatens jobs.
The threat is not that AI will replace all workers (in the short term anyway), the threat is it will increase productivity rapidly enough to replace 20%+ of workers quickly enough that new jobs won't be created fast enough to offset the losses.
I know several folks who stopped getting recruiting calls after 40ish. I bet there is a substantial reservoir of talent that's going untapped. Granted, it's totally anecdotal but I've had the discussion with at least 10 to 15 folks who are pretty awesome engineers.
Not only anecdotal, but probably complete BS as well. Unless these people are living in an area with no market for software engineers, if they are out of work for more than a few months (unless by choice) they are mediocre at best. Two of the last three employers I worked for didn't even have a position for me when I was referred to them, they created a job just to not lose out on the opportunity to hire a quality software engineer. And while I am very good at my job, I'm not some kind of one in a million rock star either.
But echoing ShanghaiBill's comment, unemployment among programmers is nearly half the overall average, so regardless of our anecdotal stories the only objective facts out there show there is a pretty big shortage of quality software engineers.
, it is expected an applicant who expects US-level wages already be a domain-specific and organisation-specific (products used) plus the applicant tracking system hell faced by anyone applying for these jobs.
So why are you worried about h1b workers or children then? If what you say is true then they arent going to be able to get those jobs either, in fact pretty much nobody is.
I added in the implied portion of his message, which left his message ambiguous if not understood by the reader.
Slashdot Mirror
Yes, years. Libratus uses 16 Terabytes of memory for just a 2 player game. The size of the game tree increases by at least a factor of 1000 when moving up to just 3 players.
This is why better algorithms are almost always a bigger factor than increased computing power when solving these problems. They won't solve playing against 3+ players with more RAM, they will solve it with better algorithms. By some cases algorithmic improvements can be 43 times more important than computing power improvements.
Considering this AI is already dealing with unknown information, I doubt the size of the "game tree" increases by the factor you cited as you add more players.
There is no sign of a death spiral in these layoffs. The article cites 1600 job openings Microsoft currently has posted on LinkedIn alone, so they are still growing. And each of these employees has 60 days to find new internal positions at Microsoft, so if their skills are still relevant they don't necessarily even need to leave Microsoft.
This is simply sign of a transitioning company that knows it cannot be successful simply sticking to the status quo.
4) Hope the job doesn't move to another country.
I believe that's a bluff and I'm prepared to call it. If they were going to do that, it'd already have been done.
What are you talking about? Offshoring of IT work already does happen, and will continue to happen. If they haven't done it already, it is because of the cost of doing it (cost in either money, effort, and/or quality). The American workforce may still benefit from limiting H1B visas, but it will almost certainly have an effect of increasing the amount of offshoring. The hope is we offshore less jobs than we save, but it's far from a guarantee.
Considering it's nearly impossible to find good software developers these days, I find it hard to believe things would be better with even less software developers in our local economy.
Well, Obama was at war for all eight years of his presidency. Beating all other presidents. Hard to argue that he was better than Bush with that record.
How is this insightful? Obama was cleaning up the mess of Bush's wars and Obama somehow has a worse record on warmongering than the guy who started them?
Put another way, 4222 US soldiers died in Iraq under Bush's watch, and 292 died under Obama's watch (that figure includes people who died in January 2009 before Obama took office). Claiming Obama had a worst war record than Bush is asinine.
My dad always said, if you have to borrow to afford something, you can't afford it.
What else was your dad wrong about?
Loans do not count. Of course you can get loans... that cannot be discharged in bankruptcy.
Why don't loans count? My ability to take out about $100k in loans with no collateral and horrible credit is directly responsible for my income rising from $40k to $160k in less than six years. I significantly turned my life around starting in 2009 (at age 28) primarily because of student loan availability regardless of credit rating or past academic history (which was also horrible for me at the time), which is only possible because they cannot be discharged in bankruptcy.
Make up your mind. Which is it? They are paying more to white males or discriminating against white males by preferring Asian workers?
Oracle has not been complying with requests for information, so the Labor Department is suing them so they can get this information during discovery. They are probably suing them for multiple offenses so they can get a more broad set of information during this process. At least that is my guess based on the information in the article.
"The brain is just a computer" is hand-wavy and objectively incorrect.
No it is not. In the context of the rest of the paragraph it is obvious he doesn't mean a digital computer with an x86 instruction set. He was referring to the brain as a machine which accepts inputs, processes the inputs based on it's current physical state, and produces various outputs to initiate action throughout the body. He mentioned Substance Dualism at the beginning of the paragraph the line you are quoting is found in order to clear up any confusion about what he meant by calling the brain a computer.
I will grant you he is using a definition of computer not used in the common English lexicon, but since we are talking about potentially future human-level AI it's likely the definition of computer in common usage could change by then to include biological computers, not just silicon.
Who uses their cell phones for phone calls anyway?
I read through the paper, but couldn't find any description of what the brain does which couldn't be considered information processing. It may not be digital processing, and it may not resemble how computers process instructions and retrieve data, but even if the physical architecture is different the brain still seems to be processing information.
He uses an example of a dollar bill, and how a person cannot recall every detail of a dollar bill from memory if asked to draw one. And that is somehow proof that the brain does not store data about the dollar bill? The person was still able to draw some details about the dollar bill, such as a person in the center and the numbers in the corners, so that data was stored somewhere. The author also makes a silly distinction between "storing data" and "changing the brain", as if the way the brain is changed isn't how it stores the data.
But neither the song nor the poem has been ‘stored’ in it. The brain has simply changed in an orderly way that now allows us to sing the song or recite the poem under certain conditions.
Sounds a lot like the brain stored the song or poem somewhere in a way it could be retrieved later, under certain conditions. Just because the brain stores information in a less precise way as a computer doesn't mean it isn't storing anything. The rest of the article continues to make similarly odd claims without backing them up. The researcher takes some very valid arguments about how many researchers rely too heavily on computer / human brain metaphors, but then he makes a lot of wild statements himself without backing them up either.
And all of those have been branches of the AI field. Since the field of AI was created, arguably during the 1956 Dartmouth workshop, it has included topics such as expert systems and statistical methods being used to make decisions based on both simple and complex input. Whether you want to accept it or not, even those Pacman ghosts' behavior can be accurately referred to as AI.
If you want a term which only refers to human level intelligence, perhaps you should use either Artificial Consciousness, Machine Consciousness, or Synthetic Consciousness. Those probably match your personal definition of AI better than the broad definition used by the scientific community.
Then you are irrelevant as far as this topic goes. Why even bother to post?
It is bad because this type of research could lead to less availability of movies on streaming services. If the studios have hard proof that Netflix is costing them money, why would they allow their movies to be shown on Netflix? Either we would see far less movies available, or the prices would go up.
Funny how Benioff mentions his Einstein feature when mentioning how much deep AI is already being used without people noticing. In this case, it would be very hard to notice since Einstein isn't even a live feature of Salesforce yet. Saying the technology is already pervasive, and then using an example that is still around the corner, is very disingenuous.
But then again, this was just Slashvertisement anyway.
That is just sensationalized reporting. The eight people in that report own $426.2 billion in assets, meaning those bottom 3.6 billion people have an average of $118 in total assets. That means the median US retiree has more wealth than 1500 "average" people in the bottom 3.6 billion poorest humans. I guess it's kind of pathetic that the average US retiree thinks they are more important than 1500 people by amassing such wealth, or some other such nonsense.
Heck, you only need a net worth of $10 to have more than the bottom 100+ million Americans combined. That is why these types of statistics are meaningless, and are only useful for grabbing headlines.
Right. Ignoring the fact that 99% of stock is owned by the 1%, so that the collective ownership of stock by the other 99% of the population amounts to approximately nothing. It's not ownership if you have no voice in its dispensation and the only thing you can do with it is sell it to someone else.
As of 2010, the wealthiest 1% of households owned 35% of all stock owned by U.S. households.
The wealth gap is bad, but not nearly as bad as you make it out to be.
I see no technology in the near term future that I think has any reasonable probability of causing mass unemployment greater than we've seen in previous generations and in previous technological eras.
Natural language processing, self-driving vehicles, and improved virtual assistants for starters. I'm not saying they are certain to cause mass unemployment, but they certainly have a reasonable probability of doing that. Job displacement caused by software create job displacement at a much faster rate than those caused by robotics, because they are deployed at a much faster rate.
So maybe we should stop calling Machine Learning 'AI'. Shall we, pretty please?
Why would we? Should we stop calling a Corvette a car? Machine learning is a subset of AI. Ever since the AI field began it involved everything from the most rudimentary rules engines to the promise of general intelligence. Only people who get most of their understanding of AI from movies think it only means Skynet.
Millions of workers still work in manufacturing.
Yes, but there are also millions of former manufacturing and other low-skill workers who cannot find work in the new economy. No one is saying everyone will be out of a job. And it doesn't even take a majority of people out of work for there to be a problem. All it takes is a small disruption to cause massive problems.
The first industrial revolution was hugely beneficial overall to workers and company owners.
Yes, eventually. But while it is easy to look at the period from about 1760 to 1840 as a small blip in history, that was eighty years where large groups of people were significantly negatively affected by changes in employment. It's also easy to look at the century where farming went from a majority of the workforce to only a few percent of us as an easy transition, without looking at how rural areas are still dealing with the loss of income and jobs today. Manufacturing came in for about half a century to help the transition, but there isn't another savior in the horizon (at least in the short term).
We already know new jobs are almost never created fast enough to help displace workers.
You can put that idea to bed by looking at employment rates.
What are you talking about? Workforce participation by working age adults is dropping fast. We are at levels not seen since the 1970's, when women participation was half what it is now. The numbers are clear an unambigious, and they point to a large portion of our country that is being displaced by technology. It is already happening. People are only worried that AI will make the problem worse, not create a new problem.
if automation lets you create twice as much value in the same time, then there's a strong argument to be made that you should get paid twice as much rather than the executives and shareholders pocketing the difference.
There isn't even a weak argument for that let alone a strong one. If investments into automation paid for by the company are causing the increase in value each employee can produce, why do the employees deserve increased pay? If the company had paid for two human assistants for each existing employee, and their quality of work improved because of it, should the employees deserve more pay then as well? Because that isn't just an analogy, it is literally the same thing as automation improving productivity.
The truth is, people don't know what they're talking about when it comes to AI. AI will be a thing in the late 22th century, perhaps. As of now, Artificial Intelligence is just a buzz word to entertain the clueless at Corporate conventions.
No one is talking about a Skynet-level general AI when they are talking about the AI which will take someone's jobs in the next few decades. They are talking about improved voice recognition, image classification, machine learning algorithms, etc. These are the technologies threatening jobs in the short term. We don't need AI robots with consciousness for workers to be displaced.
You could say that about any technology. Most of the hand wringing over AI taking everyone's jobs is the same sort of paranoid response we've had to every technology improvement. We've seen this play before. Back in the 1970s everyone was convinced industrial robots were going to take their jobs tomorrow. Robots did become an important tool but it took decades and most of the displaced workers found new employment in comparatively short order. And plenty of people are still employed on the assembly lines right next to those robots they worried about.
Talk to the millions of workers still displaced by technological advances in manufacturing about how those 1970's fears were unfounded. The only reason we aren't seeing more resistance to this problem is some politicians have convinced them outsourcing is the cause of their problems and not automation. This is a stark contrast to the actual research into their plight, which estimates the vast majority of the lost jobs — 88 percent — were taken by robots and other homegrown factors that reduce factories' need for human labor.
the threat is it will increase productivity rapidly enough to replace 20%+ of workers quickly enough that new jobs won't be created fast enough to offset the losses.
While I think your numbers are suspect, this is the only rational argument worthy of concern. There is literally an unlimited amount of work to be done but it takes some amount of time for people to adjust to new economic realities. I think that people are vastly overestimating the risks involved here but fast increases in productivity make for short term dislocations in the work force. Some people have a hard time keeping up.
In the first industrial revolution it took generations for workers to recover from crippling job losses due to new machinery. These are not short term dislocations. There may be a nearly unlimited amount of work to be done, but not an unlimited amount of work which produces enough economic value to support someone financially. I would like someone to clean my home office, for instance, but I'm not paying $15/hr for it.
No one is making predictions without strong historical and current indicators showing there is already a problem. We already know new jobs are almost never created fast enough to help displace workers. We have just been lucky enough that only small sectors of the economy has been affected at once in the past, so we can easy ignore those who suffered.
The vast majority of human history, at least 99%, was of a world where nearly everyone was in poverty by modern standards. Thinking that our current shared prosperity is somehow the status quo which can be sustained with minimal effort is ridiculous.
It shows such a lack of understanding of the problem when he says the industry should focusing on saving people time instead of replacing people. Saving workers time so they can be more efficient is what allows companies to cut staff. Saving time and working more efficiently is the whole reason AI threatens jobs.
The threat is not that AI will replace all workers (in the short term anyway), the threat is it will increase productivity rapidly enough to replace 20%+ of workers quickly enough that new jobs won't be created fast enough to offset the losses.
I know several folks who stopped getting recruiting calls after 40ish. I bet there is a substantial reservoir of talent that's going untapped. Granted, it's totally anecdotal but I've had the discussion with at least 10 to 15 folks who are pretty awesome engineers.
Not only anecdotal, but probably complete BS as well. Unless these people are living in an area with no market for software engineers, if they are out of work for more than a few months (unless by choice) they are mediocre at best. Two of the last three employers I worked for didn't even have a position for me when I was referred to them, they created a job just to not lose out on the opportunity to hire a quality software engineer. And while I am very good at my job, I'm not some kind of one in a million rock star either.
But echoing ShanghaiBill's comment, unemployment among programmers is nearly half the overall average, so regardless of our anecdotal stories the only objective facts out there show there is a pretty big shortage of quality software engineers.
, it is expected an applicant who expects US-level wages already be a domain-specific and organisation-specific (products used) plus the applicant tracking system hell faced by anyone applying for these jobs.
So why are you worried about h1b workers or children then? If what you say is true then they arent going to be able to get those jobs either, in fact pretty much nobody is.
I added in the implied portion of his message, which left his message ambiguous if not understood by the reader.