Raw LLM Responses

This is a frought talk, it's based on conjecture and speculation and distorted beliefs. "We don't know where the data came from and we don't know what it is, but it isn't licensed and it was stolen and is cometing against those iot was stolen from" ...Either you don't know, or you know it was stolen. If it was stolen, go to court, prove it, win lots of money, have the data removed from the model. If you don't know it was stolen then get off the stage or at least stop claiming that you do know it was stolen. So far my experience of AI is that it isn't that good. It's pretty easy to compete with in the things that I've used it for, so if it's competing using stolen data it should have stolen other data. It's a bit like a cat burglar breaking into a bank to steal and leaving the bank with a whole bag of pens. They're not bad, they're not without value, but they aren't as valuable as the cash that was right beside them on the counter. I don't mean to suggest that it's ok to steal as long as it's not something of primary value, but rather if they were going to steal data, if the endevours were as nefarious as Ed wants us to believe, surely they would have taken the cash not the pens. I use it primarily for IT technical reasons, I have used AI models for other things though. I've had it help me with SQL queries that I run against databases to interrogate the data for information, insights, planning information, financial information and such. It's pretty good at writing queries, but I've had to correct it numerous times. Not because my prompt was bad, but because the claims it made were as untrue as Eds claims. I've used it to configure Apache services, Apache has been around a long time, data about how to configure it is available in abundance, I have had to clean up ChatGPT's mess a few times. The same goes for things like Elasticsearch. It can be quick and convenient, but if an enterprise level sysadmin were to use ChatGPT the users and the sysadmins management would know the difference quickly, and not because chatGPT is better at it, quite the opposite. I've had an AI agent make logos for businesses (As a temporaty tide-over, never as a permanent feature or final delivery). The trouble is that AI agents don't use fonts or vectors like the local print shop does (Or the large scale printer) so Logos which are unique and fancy tend also to not be printable on everything that they need to be printed on, inevitably customers replace the AI tide over with the real thing made by actual humans. So chappo here is crying because AI learnt to run by watching Usain Bolt do it and can now run the 100m in just over a minute. It's incredible that it learnt to run that way, that it can run at all, but it isn't going to beat the Bolt. I've heard AI generated music, and fundmanetally the problem with music as an example, other than Ed is Cherry Picking, is that, like with art, how good it is is entirely subjective. You and I can hear the exact same sound and based on many things which have nothing to do with the world outside our own minds, you might enjoy it and I might not (Captain Beefheart had fans, actual fans who liked Trout Mask Replica, who genuinely enjoyed it. I think it's excellent that Trout Mask replica exists, but I'm not playing it in the car when I go to work). So for things where it's impossible to know whether something is genuinely creative, whether somehting is genuinely good, where they aren't objectively qualtifiable, an AI agen can peddle out waffle faster than Kanye West can. That said, imagine asking a human "Make music" if that human had never heard music, or alternatively having them hear only music they had paid for the license to use! One wouild have to ask ":What makes this original, is it that these sounds have never been put together in this way before or is it that it came from this person, it doesn't sound exactly like anything they have heard before although it's clear that there's a little of many things they have heard before involeved, like uniform beats, notes on a scale, the use of pre-determined instruments to make sounds which are already known? If they paid for the license, and the sound was coincidentally very similar to that oif a chart topping hit, would that be considered a creative infringement like plagiarism, or is that just lienced entitlement and legitimate use? Maybe in the future geneerative AI will be far better than a human at things like making music, it'll understand human emotions better than humans do. Ritght now it has the people skills of a left wing IT engineer. If one has knowledge of personality attributes and disorders then one can already discern some attributes which show up in personality disorder, things like putting information forward as knowledge even when it's not factual. If unlike a human, an LLM like ChatGPT has no need to impress, no need to seem competent even in those circumstances when it isn't, then why not preface information like "I'm not 100% sure of this answer". People with NPD and BPD (Any cluster B disorder really, and others) will often "want to be seen as the person who knows, who is informed and knowledgable and for that reason valuable" and will often make claims that information they have is true with no knowledge of whether or not it is in fact true. AI does the same, it'll present a configuration as though it is the final configuration with no indication that it isn't. Abstract concepts which haven't been described in data are beyond it. Ask it to describe octopus gloves and see whether it describes gloves made of octopus or whether it describes gloves worn by an octopus or whether it asumes that "octopus" in this case is a brand name, for us this isn't a stretch, we have intelligence. AI doesn't, or at least if it does it's a very very narrow defintion, one tiny subsection of what qualifies as intelligence (And if one knows how AI works the debate against intelligence would be a fairly easy one to make the point that it isn't intelligence, in). This "TED Talk" is the same kind of left wing rubbish that has people believing that entire races of humans are either good or bad, or that there's a good kind of racism based on which race it's against. A chap who deoesn't appear to understand the thing he's built an income trying to regulate the use of. I'd say this is an infomercial, not a ted talk. "Are you tired of eggs rolling under fridge and making your wife cry whenever you try to slice them, we have an egg holding device which also slices them, look ho easy it is to pack away", "Tired of all that back and neck pain you get from squeezing lemon juice out of half lemons, we have a patended device which will help you squeez a platci thing holding a lemon half instead of all that difficulty of just squeezing the lemon." "Tired of AI competing against you for your creativeity, we have a service which will allow us to take away from AI the thing that makes it make your life easier, for just a few thousand dollars. Instead of paying the AI company, pay us".