As an A.I. language model, I’d like to see a T-1000 revision 2 that can actually do it’s fucking job and kill John Connor.
An AI that can determine whether or not a given program will halt.
That’s easy:
getRandomBoolean()Sure, it might be wrong sometimes, but we can just blame it on the September 2021 training cutoff.
Something with enough context to write sensible test cases for a large codebase. It would be great if you could write test cases for a couple of domains, then ask it to write cases for a third domain following the same general style as the first. It would ideally have a conversation about what things to mock/stub and what things to keep.
I personally think 5 years isn’t enough time to get to that point with something that works really well. It’s tricky enough to get a junior up to speed with doing it sensibly, but cutting down on the time it takes to build a good test suite would mean we Devs can spend a lot more time on features and improvements.
I hope there will be AI audiobooks of this quality without the need for a team of people working on them:
I personally hope that all of this AI stuff will become a flop and disappear soon. It’s unreliable, not eco-friendly and dangerous for privacy
It really seems overblown to me. Maximum hype with minimum actual results delivered. Not saying it is useless, but I don’t think we’re going to see entire professions eliminated overnight as is being predicted.
I really think calling it “artificial intelligence” is a stretch. It’s guided pattern recognition which can do some flashy things but is not intelligence any more than parroting speech is.
Plenty of sizzle, very little beef. What will be the next flavor du jour?
NFTs, wearable tech, VR, 3D, etc…
Beanie Babies, Pogs, Tulips…
Wearable tech and VR do make some sense. NFT is not eco-friendly cuz based on blockchain. 3D? Idk it’s just for fun I guess. But at least it’s not consuming the power of a small country to run the servers
Seeing this comment upvoted so much gives me hope for humanity.
Seeing how completely crazy people across vast swathes of society have gone for AI/LLMs has significantly downgraded my opinion of other people, which was already pretty low.
You don’t have to have a wide knowledge base to figure out these things hallucinate constantly if you ask them anything that isn’t super easy to find yourself on a search engine—and if you can do that, why bother with LLM? And yet I still (with horror) see some people say that they ‘learn’ things from it and that it’s their favorite way to ‘learn’! If you aren’t constantly fact checking you’re doing the digital equivalent of casting bones to see what it says.
I hope to have something running locally that can easily sort my photos. Especially something that can handle children and non-white people. So far everything I’ve tried seems to fall short.
An ai that poisons any trace of art or text scraped by previous ai in all of its non-consent architecture that’s being propped up much the same way nfts were as “the future”
I want an LLM integration for my work’s ticketing system so that I don’t have to read the notes, I can just get a summary and ask questions about it as needed.
Virtual Assistants that can parse a command even if it has a long uhhhhh in the middle of it.
I’ve been using Google’s Gemini and it’s pretty good at interpreting fucked up or imperfect smart commands. For example we have some lights named “Chrimas Lights” and it will turn those on and off by referring to them as Christmas lights. It can also do multiple commands in a row without being overly explicit. So you can say “set lights to x%, make them yellow, and turn them off in an hour and set my TV to volume x” and it’ll do it no problem. The old assistant could not do anything even close to this.
It’s also much faster and processes words as fast if not faster than a human can. From finishing a command to the command being executed seems to be about 1/10th of a second which makes me wonder if it’s doing any sort of inferencing on the back end. It’s one of the best LLM integrations I’ve seen so far.
Or if you correct yourself in it!
I want to see games with AI-powered NPCs making responsive, possibly unique dialog. Probably text-only at first, but maybe with AI voice-acting later on.
That would blur the line between “intended behavior” and “glitch” so much.
Like, imagine NPCs breaking the fourth wall.
There are mods for this in skyrim already, and there is an AI text-based adventure game thats been around for a long time.
Maybe I’m just jaded, but at this point I’d be happy with an AI feature that would parse the appointment reminders by email or SMS from my dentist and offer to add them to my calendar with one tap.
A digital assistant that can follow me around using the data my phone provides, and then use that data to actually suggest ways to improve my life.
“My phone” knows everything about me, where I go, what stores I visit, how long I’m there, transactions, etc. It knows all my contacts, all my social relationships, all my appointments, shopping lists, all my text messages, etc.
“Assistant has determined you stop for cigars about every 4 days. We’ve found this online store that will deliver your brand at a lower cost”
“Assistant overheard you mention that you need sugar, and it is absent from your shopping list.”
"Assistant will suspend realtime driving directions because it overheard you say ‘shut the fuck up, I’m stopping at 7-Eleven first’ "
I would like to be able to take compiled machine code and generate human-readable source code.
Imagine being able to fully understand and even modify all closed-source, commercial software.
By a twitch streamer vtuber: https://github.com/cyberkaida/reverse-engineering-assistant
An AI assistant that hooks into Ghidra, explaining what things do.
You can kind of do this now. The problem is that variable names and debugger info is lost, so you wind up with all variables and functions being named v1, v2, v3, etc and f1(), f2(), f3(), etc.
Decompilation is totally a thing
Right but I assume the parent meant that the AI could guess what the original function and variable names were based on the code, and may be even clean the code up by deoptimizing it too.
Trying to make sense of decompiled code is pretty hard for any large code base.
Deoptimization could definitely happen via AI or even just really good decompilers (much like the black magic that goes into compiler optimizations), but I think that the problem with coming up with original function names is that the information just isnt there. You or the AI can make assumptions and poor guesses, but from a data theory perspective, i think that the compilation process is highly lossy. As such, the AI would essentially have to pull things out of its metaphorical ass to make anything remotely functional. I’ve had AI write me code with functions that literally don’t exist in libraries before instead of telling me it’s not possible, and I’m sure many of you reading this have too. Extrapolating that kind of behavior onto this problem, I am doubtful of a successful implementation of this half of the idea.
It sure as shit would be neat though if it could be done, and it would be a strong reciprocal to what Microsoft does with code theft to use as training data for copilot.
Animal Crossing with AI villagers
Ceiling projected GFE bot that can mesh her avatar with a real prostitute that kinda looks like my ex
