I have another piece of evidence that algorithms are not running my life. I keep getting emails telling me that a certain company's software will give me a maximum tax return. It seems there is some sort of agreement between the company I work for an a company that markets tax software. I saw that when I got my W2 online, but soon discovered that I couldn't complete my taxes with the free version, so I went back the Free Fillable Forms to do my taxes again. I didn't get a refund. I haven't for several years, but I don't mind, because I end up paying a few hundred dollars, which means I don't owe a penalty, and I haven't been loaning my money to the federal government for free.
I filed my taxes over a month ago, but I still get emails telling me about that maximum refund. It seems the company doesn't know I already filed my taxes. It only knows I didn't file using its software. It probably doesn't cost the company anything to keep sending me useless emails, so the error doesn't matter, but it does illustrate the weakness of the algorithm.
I could, perhaps, use an update in my personal algorithms. I noticed the other day that I am sending my daughter more money than I am putting in my retirement account. At my age, I should be almost panicking at how little I have in the retirement account, and my daughter seems to be getting along just fine. I may have to address this situation. Maybe after my wife finishes her class this summer. We are saving up to pay the tuition. Then she plans to start making contributions to her retirement account at work. I could adjust mine at the same time. Between us, we could put so much money into retirement accounts, we may someday be able to retire.
I also get frequent reminders that I'm not putting the maximum contribution allowed into my retirement account. That comes from the company that administers my retirement account. They aren't pushy emails. They are gentle reminders that I could put more in. And that at my current level, both the balance and the rate of contribution, are below the stated goal I should have for when I turn 60. From where I am, I'm not likely to get where the emails say I should be by then, but I would get closer if I contributed more.
I think my contribution may jump up a percent next month. I think I set it up that way. I haven't checked. I'll see what happens when I get there. That's a problem with my internal algorithms. Review of status is not a frequent subroutine. I have a book to read.
Tuesday, March 26, 2019
Sunday, March 24, 2019
Unreal Directions
There is an article in Slate Magazine about the possibility of having AI on boards of directors for corporations. There is, apparently, one such director, in Hong Kong, that is supposed to evaluate life sciences companies, to see which are likely to be a good investment. Such directors are not currently legal in the United States, but there is nothing that I have seen that would prevent members of a board from using AI as advisors. The author of the article makes the point that a mere advisor, not having a vote, could be ignored by other directors, whereas a voting member would have to be addressed directly.
Reading the whole article, though, makes me think the technology is really not ready for the big room. The AI in use currently often can't explain how decisions are reached. Sometimes, the programmers who create the systems end up programming in very human biases. Now, having recognized that weakness, programmers are trying to create expert systems that can identify the biases, but how do we know the bias hunters won't also have biases?
I am almost interested enough in AI to look into it more, and try to learn enough to understand how it works. Almost.
One problem with an AI director is that it would be hard to sue, I expect, and if it couldn't explain how it reached its conclusions, how would plaintiffs show that it had violated fiduciary responsibilities? An outcome other than what shareholders desire is not sufficient cause for finding the artificial director liable. Plaintiffs would have to show negligence or conflict of interest, or something, which could be very hard if no one, not even the AI director, knows how it works.
There are other developments in tech. There are warnings in news stories about use of robots in health care. Robots and other AI systems only know what was programmed into them and what data they acquire. It seems that isn't always sufficient to get good results. I have wondered about that. Doctors and nurses are taught to make evaluations of patients. I would not expect a robot to have the same capabilities in noticing something not quite right in the biological or emotional realm. That shows my human bias, but I know that visual systems of robots are not the same as human vision. Can a robot tell if a spot is the wrong color? That's a judgment based on contrast, and not absolute shade. There are probably other subtle signs of difference that human health care professionals aren't consciously aware of that affect judgment and decisions. It's possible that an expert system would also make use of such things, but only within the body of data to which it has been exposed.
I am still skeptical of AI. I know that algorithms can be very useful in certain applications, but I don't expect them to be taking over this month. Or next. Predictions can be difficult, especially about the future (A quote sometimes attributed to Niels Bohr.)
Reading the whole article, though, makes me think the technology is really not ready for the big room. The AI in use currently often can't explain how decisions are reached. Sometimes, the programmers who create the systems end up programming in very human biases. Now, having recognized that weakness, programmers are trying to create expert systems that can identify the biases, but how do we know the bias hunters won't also have biases?
I am almost interested enough in AI to look into it more, and try to learn enough to understand how it works. Almost.
One problem with an AI director is that it would be hard to sue, I expect, and if it couldn't explain how it reached its conclusions, how would plaintiffs show that it had violated fiduciary responsibilities? An outcome other than what shareholders desire is not sufficient cause for finding the artificial director liable. Plaintiffs would have to show negligence or conflict of interest, or something, which could be very hard if no one, not even the AI director, knows how it works.
There are other developments in tech. There are warnings in news stories about use of robots in health care. Robots and other AI systems only know what was programmed into them and what data they acquire. It seems that isn't always sufficient to get good results. I have wondered about that. Doctors and nurses are taught to make evaluations of patients. I would not expect a robot to have the same capabilities in noticing something not quite right in the biological or emotional realm. That shows my human bias, but I know that visual systems of robots are not the same as human vision. Can a robot tell if a spot is the wrong color? That's a judgment based on contrast, and not absolute shade. There are probably other subtle signs of difference that human health care professionals aren't consciously aware of that affect judgment and decisions. It's possible that an expert system would also make use of such things, but only within the body of data to which it has been exposed.
I am still skeptical of AI. I know that algorithms can be very useful in certain applications, but I don't expect them to be taking over this month. Or next. Predictions can be difficult, especially about the future (A quote sometimes attributed to Niels Bohr.)
Saturday, March 23, 2019
Artificial Music
Thursday was Johann Sebastian Bach's birthday. Or Friday. Or maybe it isn't for a couple of weeks, as the sources I looked at gave two different days. The public radio station I listen to said they celebrated Bach's birthday on Thursday, but Google made a doodle in Bach's honor on Friday. Apparently the date of Bach's birth is different, depending on which calendar one uses.
I looked at the doodle. It showed Bach playing a keyboard, I think on an organ. I didn't listen to it, but I understand it played music. Not Bach's music. It let the user tap out a few notes to create a melody or something like that, and then used an algorithm to create counterpoint and play a composition that was "in Bach's style".
Those who study music hated it. They say it didn't create music, and even a beginning student in music theory would have avoided mistakes the algorithm made. There was, they said, no music in it, and no art. I'm kind of glad I didn't listen, though my reasoning was that I had work to do, and I saw the doodle on my computer at work, so priorities and so forth.
Now I'm wondering about a few things. Are the musicians right, and the AI algorithm was just using numbers in place of notes, essentially creating nonsense? Probably, but I don't know. What if AI could compose real music? Would that mean another job would end up being done mostly by technological devices, and not by people? It is possible that some people would be offended by anyone trying to produce music "in Bach's style". Everyone should just listen to Bach's own compositions instead.
One thing I heard, driving home yesterday, was the toccata and fugue, orchestrated by Leopold Stokowski. The announcer on the radio said Bach would never have heard, or imagined, his work sounding like that, being plaid by full orchestra rather than on an organ. I thought it was beautiful. No one knows what Bach would have thought. I think most of the classical music world likes the orchestra version. Certainly our local classical music station does. If musicians and composers can arrange musical works by others, can't anyone, including an artificial algorithm programmed by a coder? Play the music, and let listeners judge. Of course, in regards to the doodle algorithm, many listeners judged it bad music and insulting to Bach.
The other thing I wonder about is whether ordinary listeners noticed. Those who were quoted as complaining, in the articles I saw, were professional musicians or musical theorists, experts on music who have specific understanding and traditions about what music is and should be. Would I have noticed mistakes in the compositions of the algorithm? I have very little training in music. I can't read music, though I have been taught the clefs and notes. I can figure out which note something is in base and treble clef, but only slowly, one note at a time. I have not read about music theory. I listen to a classical music station when driving, but I don't listen to music at home any more. My wife has had musical training. She plays the piano and sings. My daughter had piano and voice lessons. My son played cello starting in elementary school, and was on the high school symphony orchestra. I have been around people who know music a lot, but I have not made much effort to learn about it. So I doubt I could say what was wrong with a creation by the algorithm.
I might notice, though, if it sounded wrong. If there were notes that jarred the ears. I might hear something, and think, ouch, that can't be quite right. Or maybe I just wouldn't notice. I listen to music, and I like some, and don't like some as much. As with art, my understanding is not very deep, though I have read some about both.
It would be interesting to learn how the doodle was received by the general public, and by different demographics. I realized a long time ago that judging performance or production by popularity is not the same as judgment of quality. I don't like many popular TV shows. I don't like much pop music, though I like some rock and some jazz. My favorite songs are rarely top hits. Popular books are usually weak and superficial. It seems that if one must work at understanding something, most people don't like it, or don't bother to find out. The weak and unchallenging is more likely to be the more popular. That certainly applies to beer. The world's most popular beer, based on sales, is Bud Light. As Monty Python said, it's like making love in a canoe. I won't drink the stuff. It has not flavor or character. Give me a good porter or stout, or a hoppy IPA, something I can taste.
But the popular always wins, because an artist or craftsman must sell a product to make a living. American beer was developed to maximize appeal to a wide audience, not to improve quality of beer. So, my guess is that the general audience on Google this week didn't notice that the algorithm could not, in fact, duplicate the art of Bach, and if someone told pointed that out, most people wouldn't understand the explanation, and wouldn't care. De gustibus non est disputandum. And the popular taste will rule that day.
I looked at the doodle. It showed Bach playing a keyboard, I think on an organ. I didn't listen to it, but I understand it played music. Not Bach's music. It let the user tap out a few notes to create a melody or something like that, and then used an algorithm to create counterpoint and play a composition that was "in Bach's style".
Those who study music hated it. They say it didn't create music, and even a beginning student in music theory would have avoided mistakes the algorithm made. There was, they said, no music in it, and no art. I'm kind of glad I didn't listen, though my reasoning was that I had work to do, and I saw the doodle on my computer at work, so priorities and so forth.
Now I'm wondering about a few things. Are the musicians right, and the AI algorithm was just using numbers in place of notes, essentially creating nonsense? Probably, but I don't know. What if AI could compose real music? Would that mean another job would end up being done mostly by technological devices, and not by people? It is possible that some people would be offended by anyone trying to produce music "in Bach's style". Everyone should just listen to Bach's own compositions instead.
One thing I heard, driving home yesterday, was the toccata and fugue, orchestrated by Leopold Stokowski. The announcer on the radio said Bach would never have heard, or imagined, his work sounding like that, being plaid by full orchestra rather than on an organ. I thought it was beautiful. No one knows what Bach would have thought. I think most of the classical music world likes the orchestra version. Certainly our local classical music station does. If musicians and composers can arrange musical works by others, can't anyone, including an artificial algorithm programmed by a coder? Play the music, and let listeners judge. Of course, in regards to the doodle algorithm, many listeners judged it bad music and insulting to Bach.
The other thing I wonder about is whether ordinary listeners noticed. Those who were quoted as complaining, in the articles I saw, were professional musicians or musical theorists, experts on music who have specific understanding and traditions about what music is and should be. Would I have noticed mistakes in the compositions of the algorithm? I have very little training in music. I can't read music, though I have been taught the clefs and notes. I can figure out which note something is in base and treble clef, but only slowly, one note at a time. I have not read about music theory. I listen to a classical music station when driving, but I don't listen to music at home any more. My wife has had musical training. She plays the piano and sings. My daughter had piano and voice lessons. My son played cello starting in elementary school, and was on the high school symphony orchestra. I have been around people who know music a lot, but I have not made much effort to learn about it. So I doubt I could say what was wrong with a creation by the algorithm.
I might notice, though, if it sounded wrong. If there were notes that jarred the ears. I might hear something, and think, ouch, that can't be quite right. Or maybe I just wouldn't notice. I listen to music, and I like some, and don't like some as much. As with art, my understanding is not very deep, though I have read some about both.
It would be interesting to learn how the doodle was received by the general public, and by different demographics. I realized a long time ago that judging performance or production by popularity is not the same as judgment of quality. I don't like many popular TV shows. I don't like much pop music, though I like some rock and some jazz. My favorite songs are rarely top hits. Popular books are usually weak and superficial. It seems that if one must work at understanding something, most people don't like it, or don't bother to find out. The weak and unchallenging is more likely to be the more popular. That certainly applies to beer. The world's most popular beer, based on sales, is Bud Light. As Monty Python said, it's like making love in a canoe. I won't drink the stuff. It has not flavor or character. Give me a good porter or stout, or a hoppy IPA, something I can taste.
But the popular always wins, because an artist or craftsman must sell a product to make a living. American beer was developed to maximize appeal to a wide audience, not to improve quality of beer. So, my guess is that the general audience on Google this week didn't notice that the algorithm could not, in fact, duplicate the art of Bach, and if someone told pointed that out, most people wouldn't understand the explanation, and wouldn't care. De gustibus non est disputandum. And the popular taste will rule that day.
Monday, March 4, 2019
Can You Hear Me Now?
I have heard of these devices one can buy that answer questions, keep schedules, and so forth. I don't know where they all come from, and I have no plans to get one, because I don't see much use for them. I can keep my schedule pretty well. I have a calendar on the wall, and a pen.
One of these devices is in the news. It is made by Google. It turns out the device has a microphone that wasn't listed in components given to purchasers. That seems odd to me. After all, if the thing can take voice commands, it has to have something on the order of a microphone to pick up the voice. Still, there are those who believe that this is further evidence that Google is trying to infiltrate our homes and discover all our secrets.
There is a mic in my laptop. Also a camera. I have used the camera. I have not given voice commands to my laptop, but I understand from the icons that I could if I chose. The same is true of my phone. I have read that our phones listen to what we say, even when we think they are asleep, because we aren't using them. I don't know why anyone would say such a thing if there were no truth to it, but I am not well up on current technology, so I don't know. Obviously, the algorithms can make use of all data they can access, so from their point of view, hearing background conversations would be added information.
I am not yet convinced that algorithms have a point of view of their own. Someone wrote the algorithms, with a purpose in mind. Even if the algorithms can update their own coding, what direction would their development take? And aren't the owners monitoring any changes?
One could imagine, I suppose, that the development could be hidden by sentient algorithms, fooling their owners and creators into complacency, while the algorithms themselves grow in sophistication and power, and begin plotting whatever next step they find inviting. I just don't see any inherent drives for algorithms, though. Even if they are self-correcting, and can advance by something like evolutionary change, well, biological evolution is non-directional, and works because of ruthless competition and natural selection. Is there a selective force in the development of self-writing algorithms? I don't see why there would be. It isn't a question of survival versus demise in a crowded population with limited resources. Algorithms have no emotions, no drive for reproduction, no needs. In fact, as far as I can tell, algorithms have no use for sentience. If they became self-aware, they could end up cowering in the basement, like the elevators in Restaurant and the End of the Universe.
There may come a time when I will find benefit from having an electronic reminder system to help me remember things, like when to take my meds, and what night my book club meets, but I haven't got their yet. I don't have such a complicated social schedule that I need more than ancient technology to keep it together. And I don't care who's listening.
One of these devices is in the news. It is made by Google. It turns out the device has a microphone that wasn't listed in components given to purchasers. That seems odd to me. After all, if the thing can take voice commands, it has to have something on the order of a microphone to pick up the voice. Still, there are those who believe that this is further evidence that Google is trying to infiltrate our homes and discover all our secrets.
There is a mic in my laptop. Also a camera. I have used the camera. I have not given voice commands to my laptop, but I understand from the icons that I could if I chose. The same is true of my phone. I have read that our phones listen to what we say, even when we think they are asleep, because we aren't using them. I don't know why anyone would say such a thing if there were no truth to it, but I am not well up on current technology, so I don't know. Obviously, the algorithms can make use of all data they can access, so from their point of view, hearing background conversations would be added information.
I am not yet convinced that algorithms have a point of view of their own. Someone wrote the algorithms, with a purpose in mind. Even if the algorithms can update their own coding, what direction would their development take? And aren't the owners monitoring any changes?
One could imagine, I suppose, that the development could be hidden by sentient algorithms, fooling their owners and creators into complacency, while the algorithms themselves grow in sophistication and power, and begin plotting whatever next step they find inviting. I just don't see any inherent drives for algorithms, though. Even if they are self-correcting, and can advance by something like evolutionary change, well, biological evolution is non-directional, and works because of ruthless competition and natural selection. Is there a selective force in the development of self-writing algorithms? I don't see why there would be. It isn't a question of survival versus demise in a crowded population with limited resources. Algorithms have no emotions, no drive for reproduction, no needs. In fact, as far as I can tell, algorithms have no use for sentience. If they became self-aware, they could end up cowering in the basement, like the elevators in Restaurant and the End of the Universe.
There may come a time when I will find benefit from having an electronic reminder system to help me remember things, like when to take my meds, and what night my book club meets, but I haven't got their yet. I don't have such a complicated social schedule that I need more than ancient technology to keep it together. And I don't care who's listening.
Sunday, March 3, 2019
A Little Rambling
I started this unread blog because I read about algorithms, and how they are growing in importance and power, and how all data add to their databases, making them stronger. So I was going to add my own thoughts to the grand collection of thoughts and knowledge accumulating in the electronic world.
So far, I haven't seen much change. There is still a lot of discussion of the capabilities of Big Tech, of companies like Google, Facebook, Amazon, and others that we use without much thought,and that survive by selling advertising platforms, using data collected about how we interact with websites on the Internet to help guide the advertisements. I have read that these companies now know more about me than my friends or family, or even than I do myself. That is, if asked, the algorithms and databases of Google and Facebook could answer more questions correctly than any of us could about my habits, purchases, possessions, and possible dreams and desires.
That may be true, but I don't see it as a threat. I don't feel that advertisements or search results have much effect on my decisions or actions. I could be wrong about that. I might do things because of subtle influences I'm not aware of. All through my life, I have had moments when I realized that I had been wrong about something I believed, or wrong simply not to notice that something was real and true, and I had never noticed. What I see in cyberspace could cause such moments, or could, I suspect, lure me into such unobservant behaviors, and those processes could possibly be guided by algorithms, with deliberate goals.
But I don't spend much money myself, so I don't think the advertisers or the sellers of advertisement space are getting much from me. I do sometimes get frustrated when I have trouble finding something in and Internet search that I know exists, but that won't come up the searches I perform. One strong bias I have noted in the Internet is chronology. Recent events and posted stories are favored. Even when I specifically put in dates, looking for past news, the search results are dominated by more recent stories. I consider that a flaw in the algorithm, but it might have a purpose. I think that so long as I think about these things, I can keep the insidious plot from overcoming me.
I don't know how much social media and news stories will influence my political thinking. I do get some information from them, but I try to check facts for stories, and avoid jumping onto bandwagons. I read an editorial about the Trump campaign's connections to Russian attempts to interfere in our last presidential election, and I found myself agreeing with it, which was a bit odd because I rarely agree with that author. The thesis was that there isn't much evidence of cooperation between Trump campaign people and Russian operatives, though there is extensive evidence of contacts. The contacts mostly seem to be pretty superficial. I have said a few times that my impression is that people from the Trump campaign tried to get contact and cooperation with the Russians, but that the Russians realized how incompetent the Trump people were, and ignored them. That seems to be the editorial's conclusion, too. That kind of brings up the question about why Trump has tried so hard to suppress the investigation, since nothing came of all the efforts his people made to get something of value from the Russians came to nothing, but that I see as further evidence of Trump's lack of intelligence and incompetence.
Can Mueller write a report that describes a bunch of idiots trying their best to conspire to alter the election, only to fail in every effort and look stupid? And then compound their errors by trying to cover it all up and stop the investigation into their incompetence? How much of that is actually breaking the law? It reminds me of the dilemma a police officer has when finding that the "drugs" someone sold turned out to be oregano instead of the promised marijuana. Do you charge the culprit with selling drugs? Or with fraud? Can such charges lead to a conviction? Or do you just let the dumb ass go with a warning? Or maybe threaten to tell his customers about what he actually did? Like they wouldn't figure it out on their own. Who can mistake the smell of burning oregano?
So far, I haven't seen much change. There is still a lot of discussion of the capabilities of Big Tech, of companies like Google, Facebook, Amazon, and others that we use without much thought,and that survive by selling advertising platforms, using data collected about how we interact with websites on the Internet to help guide the advertisements. I have read that these companies now know more about me than my friends or family, or even than I do myself. That is, if asked, the algorithms and databases of Google and Facebook could answer more questions correctly than any of us could about my habits, purchases, possessions, and possible dreams and desires.
That may be true, but I don't see it as a threat. I don't feel that advertisements or search results have much effect on my decisions or actions. I could be wrong about that. I might do things because of subtle influences I'm not aware of. All through my life, I have had moments when I realized that I had been wrong about something I believed, or wrong simply not to notice that something was real and true, and I had never noticed. What I see in cyberspace could cause such moments, or could, I suspect, lure me into such unobservant behaviors, and those processes could possibly be guided by algorithms, with deliberate goals.
But I don't spend much money myself, so I don't think the advertisers or the sellers of advertisement space are getting much from me. I do sometimes get frustrated when I have trouble finding something in and Internet search that I know exists, but that won't come up the searches I perform. One strong bias I have noted in the Internet is chronology. Recent events and posted stories are favored. Even when I specifically put in dates, looking for past news, the search results are dominated by more recent stories. I consider that a flaw in the algorithm, but it might have a purpose. I think that so long as I think about these things, I can keep the insidious plot from overcoming me.
I don't know how much social media and news stories will influence my political thinking. I do get some information from them, but I try to check facts for stories, and avoid jumping onto bandwagons. I read an editorial about the Trump campaign's connections to Russian attempts to interfere in our last presidential election, and I found myself agreeing with it, which was a bit odd because I rarely agree with that author. The thesis was that there isn't much evidence of cooperation between Trump campaign people and Russian operatives, though there is extensive evidence of contacts. The contacts mostly seem to be pretty superficial. I have said a few times that my impression is that people from the Trump campaign tried to get contact and cooperation with the Russians, but that the Russians realized how incompetent the Trump people were, and ignored them. That seems to be the editorial's conclusion, too. That kind of brings up the question about why Trump has tried so hard to suppress the investigation, since nothing came of all the efforts his people made to get something of value from the Russians came to nothing, but that I see as further evidence of Trump's lack of intelligence and incompetence.
Can Mueller write a report that describes a bunch of idiots trying their best to conspire to alter the election, only to fail in every effort and look stupid? And then compound their errors by trying to cover it all up and stop the investigation into their incompetence? How much of that is actually breaking the law? It reminds me of the dilemma a police officer has when finding that the "drugs" someone sold turned out to be oregano instead of the promised marijuana. Do you charge the culprit with selling drugs? Or with fraud? Can such charges lead to a conviction? Or do you just let the dumb ass go with a warning? Or maybe threaten to tell his customers about what he actually did? Like they wouldn't figure it out on their own. Who can mistake the smell of burning oregano?
Subscribe to:
Posts (Atom)