chatgpt-heretic

[chatgpt-heretic]

ChatGPT is actually really good

So, I'm getting quite bummed out by the rabid AI hatred that is the vast majority opinion on tumblr and reddit. For all y'all's sake, not for mine. There are a number of significant concerns around ChatGPT and it's competitors. However, the actual negative things that I see people talk about are usually not these. Instead, they're vastly exaggerated or just false statements about how ChatGPT is worse than useless. When actual valid issues get mentioned, they're discussed inacurately and out of context.

I understand that ChatGPT can be used in inappropriate ways, for inappropriate tasks, and people can get crap results. I'm also very clear that ChatGPT is extremely useful and powerful for a wide range of personal, creative and technical work. If you've not seen this, I reckon either 1) you haven't used ChatGPT in more than a year, and are thinking of a much older version 2) you don't really use chatgpt at all, because you think it's bad based on things you've heard other people say 3) you've tried using it, but you used it for the wrong things, or in the wrong way- easily done.

I'll try and do some thorough blog posts to talk through examples. But the biggest reason it all makes me a bit sad, is this: LLM's are not going away. They are already entrenching in society, and they're likely to be the biggest change since the computing revolution. Contrary to an image I just saw on my feed, which said, amongst other things, 'everyone hates them', the majority of people don't really care, and a lot of people find them useful. They're going to be a bigger and bigger part of government, business and, for many people, our personal lives. Ranting about how shit they are is disenfranchising you of any say in this future. Because they're not going to go away. If you say 'chatgpt is useless and we should not have it', no one making any decisions about chatgpt will listen to you. If you say, here are my concerns and some evidence of why they're valid, and here's some ideas about how it could be done better, or maybe here's some people who seem to be doing it the right way, lets copy them - that could actually shape the future. People's complaints about chatgpt have led to changes in the product and the user agreement, but they were complaints about actual things, not that it's just completely crap. When I use chatgpt, I regularly get offered two versions of it's response, and I rate the one I think is better - this is user research the company will use to decide how their product should work. There was an appalling update with some major changes a few weeks ago, and it got rolled back in about 2 days, because of the number of complaints. I'm not suggesting everyone should be using chatgpt, but I do think that you are the people who I'd most want influencing the product, and I do think it's an issue when so many people are stating strong negative opinions that aren't fact-based. I'll try and write about these things in more in depth posts in a vaguely coherent way, but you're welcome to comment questions, so I can try and write about things y'all are actually interested in.

[chatgpt-heretic]

Disclaimer

OpenAI is a morally objectionable company. I'm not OK with their business practises or their prioritisation of the business over ethical concerns. Copyright theft is bad. ChatGPT was trained on data it should not have been able to use for free. However, they did it, and got away with it- the results were so successful, that they will never be truly held to account for that. Considering that cat is a very long way out of the bag, they should pay a fair amount of $$ to the creator of every copywritten piece of media they used. I'm sure amazon, google, facebook and twitter have similar issues, although amazon, google and facebook will have diluted the evil in a sea of bland corporate structure and, I assume, will be operating more cautiously. And Twitter is a literal tool to spread far right propaganda, hatred and division, so it's kinda irrelevant whatever bad things they might have done around their LLM. There are many valid concerns about the uses that LLM's could be put to by companies and governments, and the effect that consumer products will have on society. The environmental impact of training and running LLM's matters. That does not mean that we should reject AI.

I'll write some blogs to explain why, at some point, and maybe I'll remember to put a link to them right here. A note on hate mail (or politely worded disagreement) If you don't use LLM's yourself, and you get your information about LLM's from things people said on social media, as opposed to a reputable, technically informed source, please have a think about whether you're basing your opinions on facts or beliefs, and consider checking whether beliefs are factually supported.