73
Without unnecessary brand name dropping, what current technology blows you away?
(self.casualconversation)
this post was submitted on 03 Jan 2025
73 points (96.2% liked)
Casual Conversation
2260 readers
412 users here now
Share a story, ask a question, or start a conversation about (almost) anything you desire. Maybe you'll make some friends in the process.
RULES (updated 01/22/25)
- Be respectful: no harassment, hate speech, bigotry, and/or trolling. To be concise, disrespect is defined by escalation.
- Encourage conversation in your OP. This means including heavily implicative subject matter when you can and also engaging in your thread when possible. You won't be punished for trying.
- Avoid controversial topics (politics or societal debates come to mind, though we are not saying not to talk about anything that resembles these). There's a guide in the protocol book offered as a mod model that can be used for that; it's vague until you realize it was made for things like the rule in question. At least four purple answers must apply to a "controversial" message for it to be allowed.
- Keep it clean and SFW: No illegal content or anything gross and inappropriate. A rule of thumb is if a recording of a conversation put on another platform would get someone a COPPA violation response, that exact exchange should be avoided when possible.
- No solicitation such as ads, promotional content, spam, surveys etc. The chart redirected to above applies to spam material as well, which is one of the reasons its wording is vague, as it applies to a few things. Again, a "spammy" message must be applicable to four purple answers before it's allowed.
- Respect privacy as well as truth: Don’t ask for or share any personal information or slander anyone. A rule of thumb is if something is enough info to go by that it "would be a copyright violation if the info was art" as another group put it, or that it alone can be used to narrow someone down to 150 physical humans (Dunbar's Number) or less, it's considered an excess breach of privacy. Slander is defined by intentional utilitarian misguidance at the expense (positive or negative) of a sentient entity. This often links back to or mixes with rule one, which implies, for example, that even something that is true can still amount to what slander is trying to achieve, and that will be looked down upon.
Casual conversation communities:
Related discussion-focused communities
founded 2 years ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
Large Language Models.
While it’s trendy to hate on them and nitpick every single flaw, I still have a vivid memory of how terrible chatbots were just a few years ago. The fact that I can now have an actual, insightful discussion with a computer still amazes me. I hope they continue to improve to the point where it’s impossible to tell them apart from a real person.
It's funny, because what makes them more like a real person is their inability to be consistent.
Like genuinely, LLMs are indeed cool, but in many ways all we've done is create a computer that is as bad at computer tasks as a human.
It lies with the same confidence that it uses to tell the truth, mostly because it can't actually make distinctions between the two.
What we have done is even worse because by and large the marketing around AI has worked and people categorically trust what a single AI will tell them after asking the question only once and if that doesn't terrify you, you aren't paying attention.
If there is a future for ai it is in "ensembles" where you ask several independently trained and developed LLMs the same question, preferably at least checking some of the LLMs for consistency/indepth correct knowledge by asking them the same question twice in two different sessions. Then you evaluate from the ensemble of answers what you think the real answer is.
Sound tedius compared to just using a search engine that works?
...and to that question I respond emphatically - Yes, very much so!
I've been working with cursor.ai lately and it's pretty mind blowing in composer mode. I understand the hate and how the tech is limited, but it's still pretty amazing at what it can do.
What is it? The link looks like a demo page for a search engine.
Sorry wrong domain. https://www.cursor.com/
It's a fork of vsc with context aware ai integration
I see a lot of hate for them on Lemmy, but I find hem quite helpful with earning how to work with computers and also summarizing general info I search for on the internet. Rather than spend 20 minutes reading thru various websites on a topic know nothing about, I can simply ask an LLM. Eve if they are known to make mistakes, I can accommodate that possibility if the stakes are high enough by looking into it further.
A programmer friend of mine uses AI constantly now in his work. Once during a D&D game he had ChatGPT write database code for a character generator or something, I forget the actual goal, just while sitting there between turns. It churned out the tedious routine crap you always have to write for tasks like that. He said he's had it refactor his own code, and it has come up with interesting approaches he wouldn't have thought of. Pretty amazing for software that essentially just reshuffles material it's already seen to create something that looks like what a human would produce, without actually understanding what it's doing.
If it's your thing 3Brown1Blue do an excellent deep dive on neural networks including LLMs that's really informative