Are you using AI - and how?

[CommodoreMid]

AI, in its current form, is just a very complicated search algorithm that uses the vast amount of information on the internet to generate its output.

If the vast majority of the internet content thought the earth was flat, Chat GPT would write you an essay about why the earth is flat. And the general public will believe it, because it's AI and so must be correct.

That's the potentially dangerous part of this whole thing.

I love my glue in my pizza sauce!

 

[Swanee]

We did a DARPA-funded project using AI in design. In that case, the data that the AI learned from was generated by models. To kickstart it we first set the inputs over a broad range and let the model run and generate results, then trained the AI on the input/output pairs that represented “good” results.

Then we let the AI take over on setting the inputs, looking at the outputs, and training itself on the good results. The AI eventually learned to generate good designs that no one had thought of. That was eye-opening. AI demonstrated an ability to tease out relationships from these complex highly coupled models that our meatware did not see.

I participated recently in an "innovation cell" for AI in COA development and wargaming.

I saw similar results to yours. A red cell with AI is a thing of beauty. And AI understands game theory way better than your average field grade staffer.

 

[MIDNJAC]

AI, in its current form, is just a very complicated search algorithm that uses the vast amount of information on the internet to generate its output.

If the vast majority of the internet content thought the earth was flat, Chat GPT would write you an essay about why the earth is flat. And the general public will believe it, because it's AI and so must be correct.

That's the potentially dangerous part of this whole thing.

You pretty much explained my fear in much more tangible terms.

 

[sevenhelmet]

You pretty much explained my fear in much more tangible terms.

That AI makes the echo chamber echo more loudly? Why, yes…

 

[Hair Warrior]

There’s definitely a market for an AI that is grounded in PhD level math, physics, chemistry, etc. Where its answers are based on fundamental truths about how the universe works, and not just counting blog posts by flat vs round earthers.

 

[hscs]

Am thinking about having AI help me rebook travel....oh wait....

 

[MIDNJAC]

That AI makes the echo chamber echo more loudly? Why, yes…

I think that was the first, greatest success of AI, exactly as the creators had envisioned

Package them all away into neat little groups and make them hate one another. They will never bring the gallows to the CEO's neighborhoods after that, they'll be too caught up committing political violence against one another

 

[number9]

There’s definitely a market for an AI that is grounded in PhD level math, physics, chemistry, etc. Where its answers are based on fundamental truths about how the universe works, and not just counting blog posts by flat vs round earthers.

Why 9.11 is larger than 9.9......incredible

I asked chartgpt which number is bigger, 9.11 or 9.9, and he actually answered that 9.11 is bigger than 9.9, which is unbelievable

community.openai.com

 

[Gatordev]

I haven't really messed with ChatGPT a whole lot, but I've found version 3.5 to be hot garbage. 4.0 was a lot more impressive (and correct) about random niche things I asked it.

I have messed around quite a bit with ElevenLabs. That is pretty amazing, even if the output seems fairly simple.

 

[sevenhelmet]

Why 9.11 is larger than 9.9......incredible

I asked chartgpt which number is bigger, 9.11 or 9.9, and he actually answered that 9.11 is bigger than 9.9, which is unbelievable

community.openai.com

The future is so fucking bright… I can’t wait to turn my brain off and have wrong numbers and bad information thrown at me all day.

Cue the tech apologists to tell me what I am misunderstanding in 3…2…1…

 

[Hair Warrior]

The future is so fucking bright… I can’t wait to turn my brain off and have wrong numbers and bad information thrown at me all day.

Cue the tech apologists to tell me what I am misunderstanding in 3…2…1…

You’re not wrong. It’s a nascent capability with flaws. It needs years of maturing.

If you need a new logo design for your startup company or beer league charity softball team, it’s perfect. If you need to make sure the O rings on the space shuttle can tolerate the ambient ground temperature at liftoff, it shouldn’t be what you rely on.

 

[sevenhelmet]

You’re not wrong. It’s a nascent capability with flaws. It needs years of maturing.

If you need a new logo design for your startup company or beer league charity softball team, it’s perfect. If you need to make sure the O rings on the space shuttle can tolerate the ambient ground temperature at liftoff, it shouldn’t be what you rely on.

That's OK. We have engineers for that, and managers to ignore them...

 

[Spekkio]

You’re not wrong. It’s a nascent capability with flaws. It needs years of maturing.

It's actually made internet search *worse.*

If you Google simple facts and are blessed with a random AI beta response, you get paragraphs of nonsense tangentially related to the topic where the information you actually seek is buried under another link (maybe).

 

[Random8145]

Why 9.11 is larger than 9.9......incredible

I asked chartgpt which number is bigger, 9.11 or 9.9, and he actually answered that 9.11 is bigger than 9.9, which is unbelievable

community.openai.com

Interesting coincidence that you post that as an example and your screen name is number9

 

[Random8145]

It's actually made internet search *worse.*

If you Google simple facts and are blessed with a random AI beta response, you get paragraphs of nonsense tangentially related to the topic where the information you actually seek is buried under another link (maybe).

Not always. I sometimes get a decently informative AI response.