kallend 1,623 #1 Posted July 1, 2022 The first author on this paper is the algorithm:https://hal.archives-ouvertes.fr/hal-03701250/document Quote Share this post Link to post Share on other sites
riggerrob 558 #2 July 1, 2022 That summary is easier to read than the next dozen legal documents written by Canadian lawyers. Quote Share this post Link to post Share on other sites
olofscience 421 #3 July 1, 2022 (edited) AI won't be taking *my* job, because it takes creativity, strategic thinking, interacting with people on a daily basis blah blah blah. Those poor people whose livelihoods got taken by AI, they can always swap their entry-level jobs with high-level AI research, right? They only can't because they don't work hard enough. If they're poor, it's because they're lazy. </sarcasm> AI will eventually be going for all jobs, mine included. The only people who will win from this, are the people who own the AIs. Capitalism will probably collapse sometime before then. (Oh, and I love technology. Seriously. It's not technology's fault people made a broken economic system!) Edited July 1, 2022 by olofscience Quote Share this post Link to post Share on other sites
JoeWeber 2,299 #4 July 1, 2022 5 minutes ago, olofscience said: AI won't be taking *my* job, because it takes creativity, strategic thinking, interacting with people on a daily basis blah blah blah. Those poor people whose livelihoods got taken by AI, they can always swap their entry-level jobs with high-level AI research, right? They only can't because they don't work hard enough. If they're poor, it's because they're lazy. </sarcasm> AI will eventually be going for all jobs, mine included. The only people who will win from this, are the people who own the AIs. Capitalism will probably collapse sometime before then. (Oh, and I love technology. Seriously. It's not technology's fault people made a broken economic system!) I just feel terrible for all of the people who will still be alive. What a tragedy. Quote Share this post Link to post Share on other sites
kallend 1,623 #5 July 1, 2022 2 hours ago, riggerrob said: That summary is easier to read than the next dozen legal documents written by Canadian lawyers. It writes in better English than most of my students did. Consistency in tenses, correct use of adverbs, etc. Quote Share this post Link to post Share on other sites
gowlerk 1,912 #6 July 2, 2022 16 hours ago, kallend said: It writes in better English than most of my students did. Consistency in tenses, correct use of adverbs, etc. Perhaps AI will finally put an end to the relentless evolution of the language. It will be so much better when older people no longer have to cope with all changes and things remain the same forever. 2 Quote Share this post Link to post Share on other sites
metalslug 31 #7 July 2, 2022 17 hours ago, JoeWeber said: I just feel terrible for all of the people who will still be alive. What a tragedy. Were you looking for the sarcasm font for this? That's not much different from the argument that moving to renewable energy will destroy a million jobs. AI is software. Software requires hardware. Hardware requires electrical power, maintenance, supporting infrastructure and a shitload of mined minerals for fabrication and maintenance. To say that machines will be supported largely by other machines is an unrealistic circular argument. I expect it will be primarily economics and politics that prevents 'dominant' AI from happening. Incidentally, to the OP; good grammar is at the easier end of AI, even some of the autocorrect apps I use (which are not large chunks of code) are really good at grammar. I've yet to hear of anything really special that's not exclusively based on raw mathematic power, much like an all-conquering chess computer is analogous to a forklift winning a weight-lifting competition. Creativity and improvisation are still valid arguments. AI only has the illusion of creativity; as many here will know even computerised 'random number' generation is not truly random and that's amongst the very simplest examples of spontaneous thought comparison. 1 Quote Share this post Link to post Share on other sites
wmw999 2,121 #8 July 2, 2022 40 minutes ago, metalslug said: AI only has the illusion of creativity; as many here will know even computerised 'random number' generation is not truly random and that's amongst the very simplest examples of spontaneous thought compariso True, but it can be a good approximation of the kind of stretching by interpolating in new ways that expands knowledge. Maybe we need to fully define creativity, now considering AI in the equation Wendy P. Quote Share this post Link to post Share on other sites
olofscience 421 #9 July 2, 2022 (edited) 52 minutes ago, metalslug said: AI only has the illusion of creativity You can say the same about people. For example, your reply was entirely predictable. 52 minutes ago, metalslug said: as many here will know even computerised 'random number' generation is not truly random and that's amongst the very simplest examples of spontaneous thought comparison. Just shows that you don't really know all the different ways computers generate random numbers. People are way, way worse at random number generation. Measurably worse. Edited July 2, 2022 by olofscience Quote Share this post Link to post Share on other sites
metalslug 31 #10 July 2, 2022 2 minutes ago, olofscience said: Just shows that you don't really know all the different ways computers generate random numbers. Based on your posts, you believe they're generated from Hollywood movie scripts. Your 'Matrix' and 'Terminator' world awaits... Quote Share this post Link to post Share on other sites
olofscience 421 #11 July 2, 2022 (edited) 1 minute ago, metalslug said: Based on your posts Really, which posts? Edited July 2, 2022 by olofscience Quote Share this post Link to post Share on other sites
olofscience 421 #12 July 2, 2022 1 hour ago, metalslug said: Incidentally, to the OP; good grammar is at the easier end of AI Have you developed any AIs? How would you know if it's easier? Quote Share this post Link to post Share on other sites
metalslug 31 #13 July 2, 2022 43 minutes ago, olofscience said: Have you developed any AIs? How would you know if it's easier? Yes, two AI algorithms, about 25 years go as part of a hobby, albeit rudimentary and neither focused on grammar. I believe grammar AI is easier from the opinions of experts in that field whose opinions I respect, some of whom I've interacted with during my career in a related field. 21 hours ago, olofscience said: AI will eventually be going for all jobs, mine included. If you truly believe that then I suspect you're duped by Hollywood fiction. Although, I concede AI will come for your job. My espresso machine has already indicated an interest. Quote Share this post Link to post Share on other sites
olofscience 421 #14 July 2, 2022 18 minutes ago, metalslug said: I believe grammar AI is easier from the opinions of experts in that field whose opinions I respect, some of whom I've interacted with during my career in a related field. Could you identify these experts? I've also developed AIs as well, that are now being used by thousands of people. I was a contributor to a popular machine learning library, but my experience with neural nets is more limited - only 90,000 neurons using the older sigmoid activation functions. But I'm currently working on a more modern one now with ReLu units. 20 minutes ago, metalslug said: Although, I concede AI will come for your job. My espresso machine has already indicated an interest. Aw, look at the little hobbyist attempting an insult I'd be glad if an AI came for my job, I've been trying to make one for years! Quote Share this post Link to post Share on other sites
olofscience 421 #15 July 2, 2022 28 minutes ago, metalslug said: Yes, two AI algorithms, about 25 years go as part of a hobby, albeit rudimentary and neither focused on grammar. Yep, definitely not Convolutional Neural Nets, Reinforcement Learning or anything remotely capable. You probably think a bunch of 'IF' statements made it an AI. You definitely can't take my job, but I'll take your espresso maker over you - I love coffee. Quote Share this post Link to post Share on other sites
JoeWeber 2,299 #16 July 2, 2022 2 hours ago, metalslug said: Were you looking for the sarcasm font for this? That's not much different from the argument that moving to renewable energy will destroy a million jobs. AI is software. Software requires hardware. Hardware requires electrical power, maintenance, supporting infrastructure and a shitload of mined minerals for fabrication and maintenance. To say that machines will be supported largely by other machines is an unrealistic circular argument. I expect it will be primarily economics and politics that prevents 'dominant' AI from happening. Incidentally, to the OP; good grammar is at the easier end of AI, even some of the autocorrect apps I use (which are not large chunks of code) are really good at grammar. I've yet to hear of anything really special that's not exclusively based on raw mathematic power, much like an all-conquering chess computer is analogous to a forklift winning a weight-lifting competition. Creativity and improvisation are still valid arguments. AI only has the illusion of creativity; as many here will know even computerised 'random number' generation is not truly random and that's amongst the very simplest examples of spontaneous thought comparison. Did you mean a humor emoticon? Don't need them, thank you. I'm not sure why you came to me for enlightenment on the topic, I barely know an algorithm from the rhythm method, unless you're one these guy's who writes a swell reply and then looks around for a place to post it. Quote Share this post Link to post Share on other sites
mistercwood 287 #17 July 3, 2022 I wonder what sort of hits you'd get if you fed that text into a plagiarism checker? Seems to be a frequently overlooked ethical and legal question with AI generated text - it doesn't pull it out of thin air, it has to learn from large training data sets. If that training data is copyrighted in any way but was just pulled from broad data-scraping practices, how badly tainted is any future output? Anyway, for my contribution re: "AI is coming for everyone's jobs"... 1 Quote Share this post Link to post Share on other sites
billvon 2,400 #18 July 3, 2022 10 hours ago, metalslug said: AI only has the illusion of creativity You could make the same argument of human authors. They learn from the same example literature as other authors; they have similar upbringings; take similar writing courses. So any "creativity" is just how author X's synapses fire differently from author Y's. But it's not really creativity - at its base it is all just synapses firing in some order. Quote Share this post Link to post Share on other sites
olofscience 421 #19 July 3, 2022 7 hours ago, billvon said: You could make the same argument of human authors. They learn from the same example literature as other authors; they have similar upbringings; take similar writing courses. So any "creativity" is just how author X's synapses fire differently from author Y's. But it's not really creativity - at its base it is all just synapses firing in some order. The arguments against AI come in two forms: 1) the "I'm/humans are special" argument, or 2) quantitative "current AI is nowhere close to doing X". The "special" argument pretty much says that there's something magic about human intelligence that AI could never do. (Usually creativity: they compare computers isolated from a random environment, to humans in a random environment to give humans the advantage) The quantitative argument is basically "current AI can only do this now, therefore in the future it will also be the same". Which ignores the reality of technological progress. And the funny thing is, you'll see a lot of these views with people in tech - more often programmers. AI researchers are more split in their opinion. 1 Quote Share this post Link to post Share on other sites
gowlerk 1,912 #20 July 3, 2022 4 hours ago, olofscience said: The arguments against AI come in two forms: 1) the "I'm/humans are special" argument, or 2) quantitative "current AI is nowhere close to doing X". AI gets written about a lot. But nothing I know of comes close to being an example of it. It's not that humans are special, it's that we are only human and we can only do human things. Current AI is not intelligence at all, it is programing created by people. AI is a fantasy and always will be. Quote Share this post Link to post Share on other sites
kallend 1,623 #21 July 3, 2022 People have a tendency to define "intelligence" in the way that supports their position. 2 Quote Share this post Link to post Share on other sites
Phil1111 910 #22 July 3, 2022 47 minutes ago, gowlerk said: ...Current AI is not intelligence at all, it is programing created by people. AI is a fantasy and always will be. "The AI software market is expected to jump 21.3% to $62.5 billion in 2022, forecasts market research firm Gartner." That number doesn't include other corporate or private firms with in house spending in the hundreds of billions. Meta/FB lost $10 billion in its AI programs last year alone. "Analyst firm IDC is predicting that US organizations will spend $120 billion a year on artificial intelligence systems by 2025. At that level, AI will see spending grow at a compound annual growth rate of 26.0% from 2021 to 2025, with the US maintaining its position from 2019 as the biggest spender on AI, accounting for more than half of worldwide spending. " Like other technologies. It will create jobs and kill jobs. Killing jobs that free up those people to move to other endeavors. Quote Share this post Link to post Share on other sites
billvon 2,400 #23 July 3, 2022 2 hours ago, gowlerk said: Current AI is not intelligence at all, it is programing created by people. That's the most common misconception about AI. AI behavior is not programmed by people. It is taught. Not all AI's learn the same thing; indeed, much of how AIs are trained are knowing when an AI is "going into the weeds" during its training, so it can be abandoned and started again. Quote Share this post Link to post Share on other sites
gowlerk 1,912 #24 July 3, 2022 2 hours ago, billvon said: 2 hours ago, billvon said: That's the most common misconception about AI. AI behavior is not programmed by people. It is taught. I understand that. The teacher is programed. Your definition of intelligence is flawed because you work in an industry that thinks it can create intelligence. Artificial intelligence is an oxymoron. What you are doing is teaching machines to adapt to the situation in a way that you want it to. Intelligence requires judgement and insight. Quote Share this post Link to post Share on other sites
billvon 2,400 #25 July 3, 2022 Just now, gowlerk said: I understand that. The teacher is programed. No, it's not. Training data sets are usually real world data. For a self driving car, those data sets are actual sensor inputs that are used for navigation for example. For OCR they are actual characters and words. Choosing the real world data is something that a person has to do. Similar to how a teacher teaches human students, the network trainer has to choose clear examples with unambiguous interpretations, and then only later move into the tougher cases. Quote Share this post Link to post Share on other sites