"It has become a pillaging of public space." The pillaging of a resource is supposed to harm it, make it scarce, and negatively affect its quality and/or abundance. Where is the harm, in this case? People have been paying for the use of original material for centuries; how is the situation of tech companies paying publishers different? It's just a matter of negotiating the conditions. Where's the ethical side of it? Individual authors should be given a right to opt out but, if they are sufficiently compensated, I can't see why they would, other than for purely ideological irrational reasons (which are obvious in Harris's references to a minuscule marginal extra energy consumption of training an LLM that also includes math data or to the restarting a nuclear power plant, as if it were a bad thing no matter what. Meanwhile, we are looking at a renaissance of this clean industry that's been tragically neglected over the last decades). It's "big tech bad, rich bad, Musk bad, profit bad" ethos; historically, it hasn't served humanity well.
Ilya's allusion to the "minuscule marginal extra energy consumption" is very flexible and can also be used to argue in favor other expressions of civic irresponsibility, like avoiding recycling or cheating on income tax. For the record:
By 2028, scientists estimate that data center energy use will be between 325 and 580 terawatt-hours per year — roughly 6.7 to 12 percent of all U.S. electricity consumption, or the amount that 30 to 53 million households currently use.
Finally someone is willing to make the argument that consolidating oligarchy is not only economically rational, which is at least plausible, but even beneficial for humanity! I am grateful for such comments, but I don't have time to respond now; I hope some of the other readers will do so while I continue to prepare my classes.
I am a mathematician who worked in publishing starting in 1974. I worked with Martin Gardner from 1974 until his death in 2010. While he and I were working on second editions of his Mathematical Games books Macmillan and Scientific American asserted claims to control this material, contrary to the arrangements set up by Gerard Piel and Dennis Flanagan when Martin wrote his columns and books. Despite substantial evidence from many directly involved, Doug Hofstadter, Ian Stewart, Jonathan Piel and others Macmillan and Holtzbrinck refused to acknowledge Martin’s rights and to pass on to him his share of the money they collected from sales they made of those rights without his knowledge. You may be sure this pattern will be followed by those selling rights to materials for training generative AI — only the damage will be different and far greater. I have lived it. Peter Renz
"It has become a pillaging of public space." The pillaging of a resource is supposed to harm it, make it scarce, and negatively affect its quality and/or abundance. Where is the harm, in this case? People have been paying for the use of original material for centuries; how is the situation of tech companies paying publishers different? It's just a matter of negotiating the conditions. Where's the ethical side of it? Individual authors should be given a right to opt out but, if they are sufficiently compensated, I can't see why they would, other than for purely ideological irrational reasons (which are obvious in Harris's references to a minuscule marginal extra energy consumption of training an LLM that also includes math data or to the restarting a nuclear power plant, as if it were a bad thing no matter what. Meanwhile, we are looking at a renaissance of this clean industry that's been tragically neglected over the last decades). It's "big tech bad, rich bad, Musk bad, profit bad" ethos; historically, it hasn't served humanity well.
Ilya's allusion to the "minuscule marginal extra energy consumption" is very flexible and can also be used to argue in favor other expressions of civic irresponsibility, like avoiding recycling or cheating on income tax. For the record:
By 2028, scientists estimate that data center energy use will be between 325 and 580 terawatt-hours per year — roughly 6.7 to 12 percent of all U.S. electricity consumption, or the amount that 30 to 53 million households currently use.
(https://www.levernews.com/biden-boosts-ai-despite-energy-dept-warning/?utm_source=newsletter-email&utm_medium=link&utm_campaign=newsletter-article)
Finally someone is willing to make the argument that consolidating oligarchy is not only economically rational, which is at least plausible, but even beneficial for humanity! I am grateful for such comments, but I don't have time to respond now; I hope some of the other readers will do so while I continue to prepare my classes.
I am a mathematician who worked in publishing starting in 1974. I worked with Martin Gardner from 1974 until his death in 2010. While he and I were working on second editions of his Mathematical Games books Macmillan and Scientific American asserted claims to control this material, contrary to the arrangements set up by Gerard Piel and Dennis Flanagan when Martin wrote his columns and books. Despite substantial evidence from many directly involved, Doug Hofstadter, Ian Stewart, Jonathan Piel and others Macmillan and Holtzbrinck refused to acknowledge Martin’s rights and to pass on to him his share of the money they collected from sales they made of those rights without his knowledge. You may be sure this pattern will be followed by those selling rights to materials for training generative AI — only the damage will be different and far greater. I have lived it. Peter Renz
I wish your story were better known. Has it been reported?