5 ms·

I can't help but be skeptical of things like this... historically speaking, humans tend to underestimate just how many things we do that require full "general i

I can't help but be skeptical of things like this... historically speaking, humans tend to underestimate just how many things we do that require full "general intelligence" to make "common-sense" decisions. Despite recent advances in machine-learning (i.e. adding more layers to neural networks, i.e. adding more columns to a matrix), we don't really have anything that even approaches an artificial general intelligence. What we do have is a suite of sophisticated statistical algorithms that basically perform regression using huge data sets. This often approximates some subset of what a general artifical intelligence could do for some use cases that mostly apply to recognizing fuzzy patterns. But it's not even close to approaching human-level intelligence. In general, modern machine learning has focused mostly on the "pattern recognition" aspect of intelligence derived from statistical algorithms, while making little significant progress with other key aspects associated with higher cognitive processes such as general reasoning, planning and creativity (apart from simplistic rule-based approaches like decision trees, or highly domain-specific problems like using massive data sets to teach a computer to play Go.) I don't doubt that something close to general AI will eventually seriously threaten human usefulness/dignity, but I question the timeline of "decades away".

geforce 10y ago
For this reason, I wouldn't be surprised if AI actually generate more work, at least for most of our lifetime. People that will supervise and correct AI errors while/after they process data, while current jobs mostly stays "until the AI is ready to actually take over". If this makes you laugh, you have probably never worked in the public sector or big corps.
- charles-salvia 10y ago
  Yeah - see my half-joke comment below
- devonkim 10y ago
  The problem is that the financial performance and massive inertia of these organizations are becoming such financial liabilities that our economies will be held back by organizational / cultural and technical debt at the same time in these organizations. I've been observing the trend for the past decade where the top tech companies are accelerating in output while most of the large institutions are falling further and further behind. Intel can keep pushing all the way to the theoretical limits of silicon for transistor size, but the economic costs including just sheer interest on fab facilities alone of getting there would bankrupt the company or hurt its credit rating. To put it into a more concrete analogy, Google is putting out autonomous cars when tons of companies are celebrating releasing the Model T or are still currently developing it with massive delays and cost overruns. Meanwhile, much of the appetite for government spending is dwindling for various reasons and this will cause a lot of strife that hurts already-slow moving organizations. Furthermore, brand recognition by the current generation of entrepreneurs for old hat technology companies is looking absolutely horrific. The next Fortune 500 CEOs 30 years from now are likely in their 30s or 40s now and they're well aware of the lack of compelling vision by so-called "technology" companies that are really just technology re-sellers and are thus unsuitable partners for increasingly tech-reliant businesses (the few that don't have such plans may be so clueless they won't matter).
Houshalter 10y ago
What percentage of jobs actually require general intelligence though? Deep learning can only do things that are relatively simple and repetitive. But that covers quite a lot. Sure, they won't be taking the jobs of programmers or doctors any time soon. But what percentage of the population can be retrained as programmers or doctors?
- charles-salvia 10y ago
  I would turn the question around and ask you: what percentage of jobs are similar enough to repetitive, algorithmic tasks? We can joke around about it, but I think most things humans do are a lot more complicated than we realize (precisely because they seem so simple to us), which is exactly why the early optimism over AI turned out to be misguided. Even a "simple" office secretary would be hard to automate, because many of these sort of jobs are not specialized at all but consist of many ad hoc tasks that range from making phone calls to negotiating with other humans to collecting debts to getting the boss coffee, etc. Many trade jobs like plumbers, electricians, etc., require way too much "common sense" type reasoning and social interactions to be fully-automated at this point or any time in the near future.
return0 10y ago
I can think of cooking as an example of a surprisingly complicated job. OTOH i can think of many supposedly "creative" jobs that are already ripe for disruption. Anything that has to do with visual arts for example, architecture, industrial design, medical, law etc. These jobs may not be fully automated, but the human supervisor will be relegated to a secondary role.
- mjmahone17 10y ago
  Visual arts and architecture are both things that seem like they'd "suffer" from induced demand: by making it significantly easier to build a unique home, we'd be lowering the cost of people getting individualized homes, which would disproportionately increase demand. Leading to more profitable work for more architects.
  - return0 10y ago
    > Leading to more profitable work for more architects. I think it will make their job easier, which means more people will compete, which means less profitable work.
  - Broken_Hippo 10y ago
    It might not actually get better for architects. After all, the job is a combination of creativity and data points (will the building stand?). The data and knowledge bit could theoretically be taken over by a computer. Suddenly, normal folks can design their own house. After all, the system would know (based on current at the time knowledge) if the building is a sturdy enough design. I think that advancement would also mean it suddenly gets easier to do home improvements as well. Most new construction would still follow fads, since humans tend to like them - look at the popularity of the McMansion, after all. A few designs from creative folks, who are not architects, would be simply beautiful or simply weird. The folks in the supply chain - or whoever owns it - would wind up being more profitable because of more demand, though. edited for an afterthought.
- chis 10y ago
  https://www.logojoy.com/ https://www.logojoy.com/ seems surprisingly close to this. If I were starting a small business I'd probably just get my logo from this type of automated service.
Chardok 10y ago
He addresses this as that AI would be a poor hunter-gatherer, but is actually quite good already at specialized low thought jobs, and there are enough of those out there to cause some economic instability. Furthermore, AI doesn't have to be 100% error free and spectacularly outperform every human, it just has to be cheaper and as-good or better than the average worker. I can think of a good number of coworkers in past jobs that are ripe for automating away - its just a matter of time when it becomes cheap enough to implement. These jobs are not high thought jobs but the ones where the rules are already set and they just follow them.
athenot 10y ago
This is true. On the other hand, our society is getting more and more complex, and tasks are getting more and more specific. At the same time, organizations are structuring themselves so that employees are hyper-specialized and are only allowed to perform the exact set of tasks that they need to. This is where AI can thrive.
lodi 10y ago
Here's the thing: let's say that AI never advances past its current level of sophistication (circa March 2017), but it does continue to get cheaper and more ubiquitous. It seems clear to me at this point that even this may already be enough to replace many agriculture jobs, transportation jobs, manufacturing jobs, retail jobs, and perhaps even some higher level jobs in healthcare and law. Even though it's "just" neural nets and fancy statistics, that already seems to be enough to put maybe a quarter, or even third, of educated people in the developed world out of work. Nevermind billions of others in developing countries.
- wahern 10y ago
  Women nearly doubled the workforce within a couple of generations, but the sky never fell. And that shift was largely precipitated by technological advances, such as the washing machine, reduced infant mortality, and birth control. In as much as doubling the workforce is similar to halving the number of jobs in pre-existing categories, I don't see the sky falling. Like with women joining the workforce, the shift to more automation will stimulate new kinds of labor demand. And it's similarly likely to make education even more important and relevant. That said, women joining the workforce is perhaps the single largest contributor to long-term stagnant wage growth. Such shifts are unlikely to prove the end of labor, but they do create significant displacement that requires serious attention. Nothing about the "free market" guarantees that the economy and labor system will adjust quickly enough within timespans relevant to maintaining the well-being of the individuals displaced. I don't really think there'll be anything particularly challenging about the next shift. In simple terms I don't see how it could be much more problematic than the addition of women to the domestic workforce immediately followed by hundreds of millions of Chinese laborers joining the global industrial economy. We just don't appreciate how disruptive those historical events were because, despite the weak U.S. response (as compared to Germany), even unassisted the labor economy did a decent job of treading water. It's quite amazing in retrospect. Theoretically we could learn from our past mistakes and not only adjust but come out ahead in terms of income growth and equality. OTOH, we could not learn from our past mistakes, and our current issues would only be compounded. On this I'm much more pessimistic. When the electorate and political culture still pine for a labor economy that is 2 or 3 epochs in the past (i.e. ancient history!), that doesn't bode well for their ability to commit to the demands of the future. Even worse is the fact that the intellectual class seems oblivious that history is merely repeating itself. The books about robots and labor literally have almost identical titles to books I read in the 1990s in college seminars discussing the implications of women, China, the information economy, and of course the rise of robots. Some of them were written in the 1980s. One famous book, "The End of Work" (Jermey Rifkin, 1994) could probably be re-published today with nary an edit. People would find it just as timely and insightful as they did then, and it'll prove just as irrelevant 25 years from now as it is today. We saw neither the dystopia nor the paradise he envisioned, but rather (as always) a slow, systematic, and largely hidden reorganization of the labor economy that became evident, if at all, only in hindsight.
norea-armozel 10y ago
I think it's more apt to say that with automation that increases in sophistication the fewer labor hours you will need to maintain it and it's successive replacements. Meaning that absent a theoretical breakthrough in AI, we'll still see a permanent dislocation of labor across all sectors over time. It's just inevitable but how we deal with it isn't.

geforce 10y ago

For this reason, I wouldn't be surprised if AI actually generate more work, at least for most of our lifetime. People that will supervise and correct AI errors while/after they process data, while current jobs mostly stays "until the AI is ready to actually take over". If this makes you laugh, you have probably never worked in the public sector or big corps.

charles-salvia 10y ago
Yeah - see my half-joke comment below
devonkim 10y ago
The problem is that the financial performance and massive inertia of these organizations are becoming such financial liabilities that our economies will be held back by organizational / cultural and technical debt at the same time in these organizations. I've been observing the trend for the past decade where the top tech companies are accelerating in output while most of the large institutions are falling further and further behind. Intel can keep pushing all the way to the theoretical limits of silicon for transistor size, but the economic costs including just sheer interest on fab facilities alone of getting there would bankrupt the company or hurt its credit rating. To put it into a more concrete analogy, Google is putting out autonomous cars when tons of companies are celebrating releasing the Model T or are still currently developing it with massive delays and cost overruns. Meanwhile, much of the appetite for government spending is dwindling for various reasons and this will cause a lot of strife that hurts already-slow moving organizations. Furthermore, brand recognition by the current generation of entrepreneurs for old hat technology companies is looking absolutely horrific. The next Fortune 500 CEOs 30 years from now are likely in their 30s or 40s now and they're well aware of the lack of compelling vision by so-called "technology" companies that are really just technology re-sellers and are thus unsuitable partners for increasingly tech-reliant businesses (the few that don't have such plans may be so clueless they won't matter).

Houshalter 10y ago

What percentage of jobs actually require general intelligence though? Deep learning can only do things that are relatively simple and repetitive. But that covers quite a lot. Sure, they won't be taking the jobs of programmers or doctors any time soon. But what percentage of the population can be retrained as programmers or doctors?

charles-salvia 10y ago
I would turn the question around and ask you: what percentage of jobs are similar enough to repetitive, algorithmic tasks? We can joke around about it, but I think most things humans do are a lot more complicated than we realize (precisely because they seem so simple to us), which is exactly why the early optimism over AI turned out to be misguided. Even a "simple" office secretary would be hard to automate, because many of these sort of jobs are not specialized at all but consist of many ad hoc tasks that range from making phone calls to negotiating with other humans to collecting debts to getting the boss coffee, etc. Many trade jobs like plumbers, electricians, etc., require way too much "common sense" type reasoning and social interactions to be fully-automated at this point or any time in the near future.

return0 10y ago

I can think of cooking as an example of a surprisingly complicated job. OTOH i can think of many supposedly "creative" jobs that are already ripe for disruption. Anything that has to do with visual arts for example, architecture, industrial design, medical, law etc. These jobs may not be fully automated, but the human supervisor will be relegated to a secondary role.

mjmahone17 10y ago
Visual arts and architecture are both things that seem like they'd "suffer" from induced demand: by making it significantly easier to build a unique home, we'd be lowering the cost of people getting individualized homes, which would disproportionately increase demand. Leading to more profitable work for more architects.
- return0 10y ago
  > Leading to more profitable work for more architects. I think it will make their job easier, which means more people will compete, which means less profitable work.
- Broken_Hippo 10y ago
  It might not actually get better for architects. After all, the job is a combination of creativity and data points (will the building stand?). The data and knowledge bit could theoretically be taken over by a computer. Suddenly, normal folks can design their own house. After all, the system would know (based on current at the time knowledge) if the building is a sturdy enough design. I think that advancement would also mean it suddenly gets easier to do home improvements as well. Most new construction would still follow fads, since humans tend to like them - look at the popularity of the McMansion, after all. A few designs from creative folks, who are not architects, would be simply beautiful or simply weird. The folks in the supply chain - or whoever owns it - would wind up being more profitable because of more demand, though. edited for an afterthought.
chis 10y ago
https://www.logojoy.com/ https://www.logojoy.com/ seems surprisingly close to this. If I were starting a small business I'd probably just get my logo from this type of automated service.

Chardok 10y ago

He addresses this as that AI would be a poor hunter-gatherer, but is actually quite good already at specialized low thought jobs, and there are enough of those out there to cause some economic instability. Furthermore, AI doesn't have to be 100% error free and spectacularly outperform every human, it just has to be cheaper and as-good or better than the average worker. I can think of a good number of coworkers in past jobs that are ripe for automating away - its just a matter of time when it becomes cheap enough to implement. These jobs are not high thought jobs but the ones where the rules are already set and they just follow them.

athenot 10y ago

This is true. On the other hand, our society is getting more and more complex, and tasks are getting more and more specific. At the same time, organizations are structuring themselves so that employees are hyper-specialized and are only allowed to perform the exact set of tasks that they need to. This is where AI can thrive.

lodi 10y ago

Here's the thing: let's say that AI never advances past its current level of sophistication (circa March 2017), but it does continue to get cheaper and more ubiquitous. It seems clear to me at this point that even this may already be enough to replace many agriculture jobs, transportation jobs, manufacturing jobs, retail jobs, and perhaps even some higher level jobs in healthcare and law. Even though it's "just" neural nets and fancy statistics, that already seems to be enough to put maybe a quarter, or even third, of educated people in the developed world out of work. Nevermind billions of others in developing countries.

wahern 10y ago
Women nearly doubled the workforce within a couple of generations, but the sky never fell. And that shift was largely precipitated by technological advances, such as the washing machine, reduced infant mortality, and birth control. In as much as doubling the workforce is similar to halving the number of jobs in pre-existing categories, I don't see the sky falling. Like with women joining the workforce, the shift to more automation will stimulate new kinds of labor demand. And it's similarly likely to make education even more important and relevant. That said, women joining the workforce is perhaps the single largest contributor to long-term stagnant wage growth. Such shifts are unlikely to prove the end of labor, but they do create significant displacement that requires serious attention. Nothing about the "free market" guarantees that the economy and labor system will adjust quickly enough within timespans relevant to maintaining the well-being of the individuals displaced. I don't really think there'll be anything particularly challenging about the next shift. In simple terms I don't see how it could be much more problematic than the addition of women to the domestic workforce immediately followed by hundreds of millions of Chinese laborers joining the global industrial economy. We just don't appreciate how disruptive those historical events were because, despite the weak U.S. response (as compared to Germany), even unassisted the labor economy did a decent job of treading water. It's quite amazing in retrospect. Theoretically we could learn from our past mistakes and not only adjust but come out ahead in terms of income growth and equality. OTOH, we could not learn from our past mistakes, and our current issues would only be compounded. On this I'm much more pessimistic. When the electorate and political culture still pine for a labor economy that is 2 or 3 epochs in the past (i.e. ancient history!), that doesn't bode well for their ability to commit to the demands of the future. Even worse is the fact that the intellectual class seems oblivious that history is merely repeating itself. The books about robots and labor literally have almost identical titles to books I read in the 1990s in college seminars discussing the implications of women, China, the information economy, and of course the rise of robots. Some of them were written in the 1980s. One famous book, "The End of Work" (Jermey Rifkin, 1994) could probably be re-published today with nary an edit. People would find it just as timely and insightful as they did then, and it'll prove just as irrelevant 25 years from now as it is today. We saw neither the dystopia nor the paradise he envisioned, but rather (as always) a slow, systematic, and largely hidden reorganization of the labor economy that became evident, if at all, only in hindsight.

norea-armozel 10y ago

I think it's more apt to say that with automation that increases in sophistication the fewer labor hours you will need to maintain it and it's successive replacements. Meaning that absent a theoretical breakthrough in AI, we'll still see a permanent dislocation of labor across all sectors over time. It's just inevitable but how we deal with it isn't.