Tweet
Re: Meet "Baxter" the Robot Out to Get Your Minimum-Wage
I'll summarize my thoughts on this interesting thread. Thank you for starting it.
The premise of the 60 Minutes coverage of Baxter the Automaton misframes important questions raised by this product. The shows producers ask, "Will robots take jobs away from workers who need them?" and "What will the workers do who are replaced by Baxter?"
Wrong questions, as usual, reinforcing ignorance in American society of the role of technology as a driver in the capitalist process to vastly improve our living standards since the Industrial Revolution and even more rapidly and profoundly since the start of the Digital Revolution.
The first issue I take with the 60 Minutes Machine Causes Unemployment formulation is the lack of diligence dedicated to understanding what Baxter can and cannot do, because only with a clear understanding of the capabilities of the machine can its social implications be discussed. That's where the story should begin, but it never gets there. It starts with the sensationalist, junk science assumption that Baxter can do anything a human worker can do.
Baxter can't see, at least in the sense of comprehending what it's looking at, or walk. A lab rat has greater capabilities in that regard. It can't talk or read. It can only perform a small number of pre-programmed activities under ideal conditions. As a tool for accomplishing work tasks it has a fraction of the capabilities of the most intellectually and physically impaired human child. But millions of viewers are taken in by its humanoid appearance and the positioning by the show's producers of the machine as contributing to unemployment, a hot button that the media can press to get an excited, emotional, and uncritical response from viewers.
The tragedy revealed by the story is that so few consumers of media have any real capacity for critical thinking. This makes them vulnerable to all manner of manipulation. It allows governments and corporations to exaggerate and lie with impunity. They entice Tweeters and bloggers to become their agents, spreading self-interested propaganda and nonsense by promising an emotional and puerile "conversation" that turns serious public policy matters into base entertainment.
Baxter, like the Google "driverless car," reminds me of "speech recognition" systems that came out from Dragon and other vendors in the 1990s. I worked for one of these companies as VP OEM Sales at the time. One day I did a live demonstration to Michael Dell and his 12 person Strategic Products group in Austin, Texas. The meeting with Dell was for consideration by him and his team to include the speech technology with every Dell PC. It was my third meeting at Dell, which is to say it took only three meetings to get to the top decision maker. I was also trying to OEM the product to HP. After nine meetings I still couldn't figure out who was in charge of making third party software decisions, which tells you everything you need to know about why HP failed in the enterprise PC market that Dell dominated. Less than a minute into the demonstration, Michael interrupted me as I recited my memorized script that appeared as words projected on the screen in the darkened room. Dell, being no dummy, had done his homework. He understood that these products, marketed as speech recognition systems, were in reality dictation recognition systems. All of them work more or less they same way.
The lead engineer once explained the technology to me in the car on the way to a meeting. They break speech down into phonemes, a unit of speech more or less analogous to a syllable. They then assemble the phonemes into a set of possible words that the phoneme groups might be. They then chose the most likely word based on grammatical rules determined by the language model for the speaker, such as standard dialect American English. British English is a completely different language model, as distinct as English from German; a brit trying to talk to an American English system will not enjoy the gobbledegook text that results. To put the word list in the context of the language model's grammar the system needs a complete sentence, but the average person does not speek in complete sentences. To speak in complete sentences is to dictate, a difficulty learned skill. Doctors and lawyers know how to dictate, which is why "speech recognition" systems succeeded in these vertical markets. But it's the broad horizontal market of his enterprise customers that Michael Dell cares about and he knows that 99% of his customers have no dictation skills. Additionally the systems have to be trained for each user and a special high quality noise-canceling microphone and headset has to be used.
He starts to give me phrases to speek into the the system. Without full sentences the word choices made by the system are erratic. The system produces nonsense sentences. Often the results are hilarious. Michael, his team, and I proceed to spend the next five minutes trying to get the most amusing results possible. "The parking lot has no time for a carrot." We have a good laugh. Dell says he thinks his customers won't have the time or patience to train the systems and won't want to wear a headset and look like "Suzy Telemarketer." The answer was no. I didn't feel too bad about it. Turned out the deal was going to be that we pay Dell to put the product on Dell computers by way of marketing to Dell enterprise customers in the hope that they later buy the product for personal use.
The legitimate name for the misnamed Speech Recognition technology is Dictation Recognition. To call it Speech Recognition is a kind of lie. It's sets a false expectation, as does Baxter marketed as a worker replacement.
Same goes for Google's so-called "Driverless Car." Once Google fesses up to the fact that the car can't be driven except on a route that was recently mapped and admits how frequently on average the system returns control to the driver because it encounters a situation that it can't cope with, then if they are honest they will rename it the Google Computer-Assisted Car, because that's what it is. Under special conditions the system can drive the car, but sometimes -- likely most of the time -- it can't.
That's were the 60 Minutes Baxter story should have started, with a skeptical eye on the claims of Baxter's manufacturer. But even if Baxter could replace humans in a wide variety of repetitive tasks, why is that bad? Humans are not designed to do the same activity over and over. It's bad for our bodies, resulting in receptive stress injuries like carpal tunnel syndrome, and for our brains, causing depression and other forms of mental illness.
Lucky to have a job? Chinese workers sit inside a metal press.
To my way of thinking asking a human to do a machine's work -- to work like a machine -- is cruel and inhuman. The right question for 60 Minutes to have asked is, How can workers who were doing a machine's job and were replaced by machines be trained for more humane employment?"
Part of the answer is in the automation itself. The Machine Causes Unemployment formulation completely misses the reality of the role of automation in a capitalist system. The reality is that a new more humane and intellectually stimulating job is created by the machine. How?
A corporation is motivated by its obligations to shareholders to increase profits and use capital in the most efficient way possible. The only reason a corporation will deploy machines to replace human workers is if it increases profitability. What does the corporation do with the additional profits that automation created? It can either distribute the profits to shareholders or reinvest the profits back into the company. If it distributes the profits then it pleases shareholders in the short run but will displease them in the long run because a competitor, seeking similar advantage, that also deploys machines but reinvests the resulting profits effectively in new product development will gain market share and competitive advantage.
The new jobs created from the profits generated from savings from automation are higher level, creative, skilled, and better paying jobs -- product R&D, management, marketing, sales. These require more training and education than the automaton jobs they replaced. This is occurring across the entire economy in all industries as machines replace tedious and receptive occupations: milking cows, inspecting circuit boards, connecting phone calls.
Here is a list of the fastest growing occupations that cannot be replaced by machines that the Bureau of Labor Statistics thinks will be in demand from 2012 to 2020. How will workers get trained for these jobs? This comprehensive study by the BLS identifies a mismatch between training, skill levels, and jobs in the future.
This is the heart of the issue raised by automation in the digital age and the one that 60 Minutes should be putting forward instead of sensationalist garbage about Baxter the Robot. But that requires that the show's producers and reporters do actual journalism, and ask viewers to actually think -- and then, armed with an understanding of the issues, vote in an informed way.
Apparently this is all bad for 60 Minutes' business, and we wonder why our country is such a mess. Baxter isn't the problem.
Originally posted by NCR85
View Post
The premise of the 60 Minutes coverage of Baxter the Automaton misframes important questions raised by this product. The shows producers ask, "Will robots take jobs away from workers who need them?" and "What will the workers do who are replaced by Baxter?"
Wrong questions, as usual, reinforcing ignorance in American society of the role of technology as a driver in the capitalist process to vastly improve our living standards since the Industrial Revolution and even more rapidly and profoundly since the start of the Digital Revolution.
The first issue I take with the 60 Minutes Machine Causes Unemployment formulation is the lack of diligence dedicated to understanding what Baxter can and cannot do, because only with a clear understanding of the capabilities of the machine can its social implications be discussed. That's where the story should begin, but it never gets there. It starts with the sensationalist, junk science assumption that Baxter can do anything a human worker can do.
Baxter can't see, at least in the sense of comprehending what it's looking at, or walk. A lab rat has greater capabilities in that regard. It can't talk or read. It can only perform a small number of pre-programmed activities under ideal conditions. As a tool for accomplishing work tasks it has a fraction of the capabilities of the most intellectually and physically impaired human child. But millions of viewers are taken in by its humanoid appearance and the positioning by the show's producers of the machine as contributing to unemployment, a hot button that the media can press to get an excited, emotional, and uncritical response from viewers.
The tragedy revealed by the story is that so few consumers of media have any real capacity for critical thinking. This makes them vulnerable to all manner of manipulation. It allows governments and corporations to exaggerate and lie with impunity. They entice Tweeters and bloggers to become their agents, spreading self-interested propaganda and nonsense by promising an emotional and puerile "conversation" that turns serious public policy matters into base entertainment.
Baxter, like the Google "driverless car," reminds me of "speech recognition" systems that came out from Dragon and other vendors in the 1990s. I worked for one of these companies as VP OEM Sales at the time. One day I did a live demonstration to Michael Dell and his 12 person Strategic Products group in Austin, Texas. The meeting with Dell was for consideration by him and his team to include the speech technology with every Dell PC. It was my third meeting at Dell, which is to say it took only three meetings to get to the top decision maker. I was also trying to OEM the product to HP. After nine meetings I still couldn't figure out who was in charge of making third party software decisions, which tells you everything you need to know about why HP failed in the enterprise PC market that Dell dominated. Less than a minute into the demonstration, Michael interrupted me as I recited my memorized script that appeared as words projected on the screen in the darkened room. Dell, being no dummy, had done his homework. He understood that these products, marketed as speech recognition systems, were in reality dictation recognition systems. All of them work more or less they same way.
The lead engineer once explained the technology to me in the car on the way to a meeting. They break speech down into phonemes, a unit of speech more or less analogous to a syllable. They then assemble the phonemes into a set of possible words that the phoneme groups might be. They then chose the most likely word based on grammatical rules determined by the language model for the speaker, such as standard dialect American English. British English is a completely different language model, as distinct as English from German; a brit trying to talk to an American English system will not enjoy the gobbledegook text that results. To put the word list in the context of the language model's grammar the system needs a complete sentence, but the average person does not speek in complete sentences. To speak in complete sentences is to dictate, a difficulty learned skill. Doctors and lawyers know how to dictate, which is why "speech recognition" systems succeeded in these vertical markets. But it's the broad horizontal market of his enterprise customers that Michael Dell cares about and he knows that 99% of his customers have no dictation skills. Additionally the systems have to be trained for each user and a special high quality noise-canceling microphone and headset has to be used.
He starts to give me phrases to speek into the the system. Without full sentences the word choices made by the system are erratic. The system produces nonsense sentences. Often the results are hilarious. Michael, his team, and I proceed to spend the next five minutes trying to get the most amusing results possible. "The parking lot has no time for a carrot." We have a good laugh. Dell says he thinks his customers won't have the time or patience to train the systems and won't want to wear a headset and look like "Suzy Telemarketer." The answer was no. I didn't feel too bad about it. Turned out the deal was going to be that we pay Dell to put the product on Dell computers by way of marketing to Dell enterprise customers in the hope that they later buy the product for personal use.
The legitimate name for the misnamed Speech Recognition technology is Dictation Recognition. To call it Speech Recognition is a kind of lie. It's sets a false expectation, as does Baxter marketed as a worker replacement.
Same goes for Google's so-called "Driverless Car." Once Google fesses up to the fact that the car can't be driven except on a route that was recently mapped and admits how frequently on average the system returns control to the driver because it encounters a situation that it can't cope with, then if they are honest they will rename it the Google Computer-Assisted Car, because that's what it is. Under special conditions the system can drive the car, but sometimes -- likely most of the time -- it can't.
That's were the 60 Minutes Baxter story should have started, with a skeptical eye on the claims of Baxter's manufacturer. But even if Baxter could replace humans in a wide variety of repetitive tasks, why is that bad? Humans are not designed to do the same activity over and over. It's bad for our bodies, resulting in receptive stress injuries like carpal tunnel syndrome, and for our brains, causing depression and other forms of mental illness.
Lucky to have a job? Chinese workers sit inside a metal press.
To my way of thinking asking a human to do a machine's work -- to work like a machine -- is cruel and inhuman. The right question for 60 Minutes to have asked is, How can workers who were doing a machine's job and were replaced by machines be trained for more humane employment?"
Part of the answer is in the automation itself. The Machine Causes Unemployment formulation completely misses the reality of the role of automation in a capitalist system. The reality is that a new more humane and intellectually stimulating job is created by the machine. How?
A corporation is motivated by its obligations to shareholders to increase profits and use capital in the most efficient way possible. The only reason a corporation will deploy machines to replace human workers is if it increases profitability. What does the corporation do with the additional profits that automation created? It can either distribute the profits to shareholders or reinvest the profits back into the company. If it distributes the profits then it pleases shareholders in the short run but will displease them in the long run because a competitor, seeking similar advantage, that also deploys machines but reinvests the resulting profits effectively in new product development will gain market share and competitive advantage.
The new jobs created from the profits generated from savings from automation are higher level, creative, skilled, and better paying jobs -- product R&D, management, marketing, sales. These require more training and education than the automaton jobs they replaced. This is occurring across the entire economy in all industries as machines replace tedious and receptive occupations: milking cows, inspecting circuit boards, connecting phone calls.
Here is a list of the fastest growing occupations that cannot be replaced by machines that the Bureau of Labor Statistics thinks will be in demand from 2012 to 2020. How will workers get trained for these jobs? This comprehensive study by the BLS identifies a mismatch between training, skill levels, and jobs in the future.
This is the heart of the issue raised by automation in the digital age and the one that 60 Minutes should be putting forward instead of sensationalist garbage about Baxter the Robot. But that requires that the show's producers and reporters do actual journalism, and ask viewers to actually think -- and then, armed with an understanding of the issues, vote in an informed way.
Apparently this is all bad for 60 Minutes' business, and we wonder why our country is such a mess. Baxter isn't the problem.
Comment