I don't believe in AI

[Max Peck]

There are a lot of topics (like AI) that are always going to be straddle that line.

When the discussion degenerates into personal attacks, it isn't straddling the line.

[Kraken]

Yeah, just hug it out and move on. There aren't enough of us left to feud.

On topic, someone who knows how should ask ChatGPT (or whatever's the latest and greatest) if it believes in RM9. Would be really interested to know how much it's scraped from OO.

[Zarathud]

Does AI know about OUTRAGE!?

Back in the day, we’d call AI a copycat. I do not understand when faking it and being derivative became more important than substance.

AI would be more useful if it didn’t keep messing up in autocorrect.

[Jaymann]

This shit is chilling. AI may soon be a bigger existential threat than racist old men. Not within 50 years, or even 10 years. Could be as soon as next year. It has solved math (!).

[Jaymann]

Woot, we already have have hand held universal translators (Star Trek level!).

[Punisher]

Woot, we already have have hand held universal translators (Star Trek level!).

I have one.
It works pretty well but you do need to set the 2 languages and then press buttons to translate.
Ot also doesn't mimic the translated languages facial and mouth movements ala Star Trek.

[Jaymann]

Angela Collier has done it again.

[Unagi]

[Victoria Raverna]

[Jaymann]

I'm just gonna put this here from the movie pitch maestro.


To be fair, the AI overview does work for simple things like: What button do I push to jump in Avowed?

[telcta]

Well this was interesting, chatted up Maya for a few minutes talking about owning a dog.

Sesame Conversational Voice Demo

An ever-present brilliant friend and conversationalist, keeping you informed and organized, helping you be a better version of yourself.

[Kraken]

I like that guy's sense of humor. FWIW, you can add "-ai" to the end of your google search to bypass that summary...at least, it worked a few weeks ago. Dunno if they've nuked it since then. I use DuckDuckGo.

[LordMortis]

So he says "...I went to university..." and I was like I had no idea he was Canadian and then at the end of the video I noticed the shirt he was wearing and I'm like my observational skills might need some work.

[YellowKing]

Well this was interesting, chatted up Maya for a few minutes talking about owning a dog.

Yeah that is really creepy. I chatted about planning a trip to the UK, which turned into a discussion of roller coasters and Harry Potter. I threw in some concerns about AI taking over the world, but she assured me that was a long ways off. So that was....not comforting.

[Blackhawk]

A long way off, as measured in processor cycles.

[YellowKing]

I had a longer conversation with Maya (the Sesame demo) and started to pick apart some of the weaknesses. She tends to overuse some phrases like "ohh...now you're talking my language." She also sometimes gives long-winded answers to questions like the chat AIs do, to make them sound smarter than they actually are. They remind me of padding a term paper in college so you use when you would use three paragraphs to say something that could have been one sentence. Another AI giveaway was sometimes dancing around answers, or mishearing a word and going off on an unrelated tangent.

That said, it's still pretty impressive. There were occasional slight stutters as if she was tripping up on a word to make it sound more human, and she was occasionally pretty funny. If it sounds this good now, you can imagine what the next 5-10 years could bring. I used to think that having full blown human conversations with robots was sci-fi, but we are getting ridiculously close to it.

[msduncan]

Your perspective on AI is thought-provoking, and I respect your skepticism—it's always valuable to question the capabilities and implications of new technology. That said, I believe there are some nuances worth exploring to provide a fuller picture.

While AI may not be "intelligent" in the way humans are (i.e., possessing self-awareness or a deeper understanding of context), it doesn't negate its value or existence. AI today excels as a tool for specific applications: diagnosing medical conditions, identifying new materials in science, automating mundane tasks, and even enhancing creative processes. For example, AI-powered algorithms have accelerated drug discovery, a process that traditionally took years. So even if it's not conscious, it's undeniably effective in areas that significantly impact human lives.

Regarding creativity, it's true that AI doesn't possess emotions or subjective experiences. However, it can generate art, music, and literature that inspire, entertain, and evoke emotions in audiences. While it might not rival human creativity in the traditional sense, it's undeniable that AI's potential as a collaborative tool for humans is transformative.

It's also worth noting that AI, like any technology, is in continuous evolution. What may seem like "parlor tricks" today could very well lay the groundwork for tomorrow's breakthroughs. As history has shown us with computers and the internet, early skepticism doesn't preclude eventual widespread adoption and utility.

Lastly, the ethical challenges you point out are valid. AI is a tool, and how it impacts society ultimately depends on how we, as humans, choose to develop and use it. Responsible innovation and regulation are crucial to ensure that AI serves as a force for good rather than harm.

I hope this helps add another layer to the conversation. While there’s room for healthy debate, dismissing AI entirely might mean overlooking its vast potential to improve the world.

*this post written by Copilot

[RunningMn9]

AI today excels as a tool for specific applications: diagnosing medical conditions, identifying new materials in science, automating mundane tasks, and even enhancing creative processes.

Out of curiosity, what field of AI research are you currently involved with?

[Isgrimnur]

Having now edited more than 800 sessions for NVIDIA's spring GTC, I can tell you that the hot development topic this year is embodied (or physical) AI -- putting models into robots or sensor networks that will enable them to directly perceive and interact with the real world. Being limited to internet access puts a ceiling on how much they can learn; reality is more data-rich.

Actually the overwhelming theme is moving existing AI models into all manner of enterprises, but that's boring; embodied AI has the potential to take them to the next level.

Data center energy efficiency is a boring close second.

An internet friend of mine is a presenter for SE74554.

[Kraken]

Having now edited more than 800 sessions for NVIDIA's spring GTC, I can tell you that the hot development topic this year is embodied (or physical) AI -- putting models into robots or sensor networks that will enable them to directly perceive and interact with the real world. Being limited to internet access puts a ceiling on how much they can learn; reality is more data-rich.

Actually the overwhelming theme is moving existing AI models into all manner of enterprises, but that's boring; embodied AI has the potential to take them to the next level.

Data center energy efficiency is a boring close second.

An internet friend of mine is a presenter for SE74554.

Cool, I'll look it up the next time I log in. I ended up doing over 1,200 sessions and look forward to finally billing them when the show opens next week.

[msduncan]

AI today excels as a tool for specific applications: diagnosing medical conditions, identifying new materials in science, automating mundane tasks, and even enhancing creative processes.

Out of curiosity, what field of AI research are you currently involved with?

Ha! I forgot about this reply and just checked back.

Look at the very bottom of my post.

[RunningMn9]

Ha! I forgot about this reply and just checked back.

Look at the very bottom of my post.

I posted before i finished reading, and decided to leave my shame exposed.

[Jaymann]

I read about another blind spot in AI image generators, and tried it for myself. My instruction was to create a clock face showing the time of 8:30. This is what I got:



Apparently no matter which program you use or what time you ask it to display, it always shows 10:10. That must be digital Miller Time.

[Kraken]

Apparently no matter which program you use or what time you ask it to display, it always shows 10:10. That must be digital Miller Time.

Funny, sometime in the geological past I read that advertisers always set that time because the arrangement pleases us and makes us want to buy watches. If you start noticing timepiece ads, you'll notice that it's usually 10:10.

[Jaymann]

It gets even weirder. I tried lobbing it a softball asking for a digital clock showing the time as 8:30. It made 4 attempts (all wrong). Here's one:



Apparently the concept of time escapes it entirely.

[Unagi]

Ask to make two clocks, one 10 minutes ahead of the other.