The AI thread

Hangar Flying
1

Where, oh where is @fearlessfrog when one feels that his insight would benefit the conversation?

Having not had time, nor a login, which was seemingly saturated every time I attempted to create an OpenAI.com account, I had not tried ChatGPT until this past weekend. My wife works for a company with AI in the name, that ostensibly uses the technology to create a dashboard to show their clients where they rate over a number of metrics against their competition. She having MBA after her name, being 14 years my junior, and being a member of the primary competing gender, I occasionally go to lengths to demonstrate why my knowledge of any technical subject superior to hers. I usually fail.

However, Saturday, while sitting with other soccer parents, a good friend of mine mentioned that he is thinking of selling his business, worried that AGI, something like ChatGPT, would make his very lucrative consultancy, somewhat irrelevant. Upon returning home, I found that OpenAI had finally found me worthy of credentials, which essentially sent me down an AI rabbit hole the rest of the weekend, the NCAA basketball tournament notwithstanding.

I’ll cut to the chase and list a couple of the very insightful videos that I consumed, and hope that this begins relevant discourse on the subject, of which I have barely scratched the surface.

Tom Scott’s shares his thoughts on the advent of AGI as it relates to the Google labels paradigm.

Lex Fridman’s interview with OpenAI’s CEO, Sam Altman.

5 Likes
2

I expect the first thing you’ll do is this


chatGPT create me a fully dynamic campaign engine for DCS, in the same vein as Falcon 4 BMS. Make it so.

7 Likes
3

My first reaction (version 3.0 -ish) was: it’s the next-level search engine.

I’ve not watched those videos yet (will shortly) but from some others I have watched (only 2, maybe 3 quick ones); first - I’m not sure ‘we’ even know what intelligence is. Sure, it’s a cool app and as far as I can tell it can be helpful and and likely convenient (akin to my I-Phone).

It may replace some things (Google ended my encyclopedia subscription - bad for the publishers) but world changing (not your words, just some I’ve heard)? Only if we let it. The general public’s first reaction isn’t always spot-on. Being adaptable is the trick.

As a test I wanted it to write me some code. I even asked for it in two languages; A few months ago I [lazily] wrote a quick recursive function (just to get something within an algorithm started). When I went back to do the non-recursive version I thought, “Hey, lets see if this thing [GPT3.5] can do that”.

If I guided it along it did fine. But it didn’t do anything more than I did decades ago when I wanted to ‘look up’ how to do this. It just “looked it up” faster. A lot faster. It’s another tool. One that can give you perhaps more time in a given context. Then what? ← that’s the real question to ask it.

Back to work :wink:

2 Likes
4

Someone on Twitter asked it what jobs it would be replacing. Clearly it had never done any of the jobs on the list and thus was a list of assumptions - unsurprisingly.

The information it provides is only good as the source like anything.

It story writing is better than mine and if you do mission descriptions it can probably come up with something.

I was hoping it would code loads of stuff for me but maybe in a few years.

Can see it being used for homework quite a bit
maybe.

2 Likes
5

From my point of view it looks like this:

This is a fancy autocomplete. It has no idea what it is saying, it only knows the words fit together. And it lies. Lies over lies, and it invents the sources if you tell it that you want sources. It has no concept of logic and it is far away from anything that I would call intelligent.

It is a decent tool for some things, like spell checking, or somewhat creative recombination of known things, like describing rooms in a dungeon (useful for me because I am a D&D Dungeon Master).

I am curious how it will develop, and yes it has great potential and risks.

5 Likes
6

This.

2 Likes
7

The running gag is, programming used to be 2h of coding, 6h of debugging. Now programming is 5min of coding, 12h of debugging.

7 Likes
8

LOL, you are really taking OpenAI to task.

Yeah, it’s not surprising that when you go to sign up for the Bard waitlist, that the site wants to tell you about the problems with large language models.

I think that Altman mentions in his interview that one of the inherent dangers is accepting every answer as the truth. Are you on the waitlist for GPT4?

2 Likes
9

My first question was to get some romantic ideas for my wife and it replied with a plethora of great activities. But I began wondering that if I do something too creative and romantic, that my wife will suspect that I’m using ChatGPT as a marriage therapist. I guess that you can’t win that one :rofl:

3 Likes
10

As a parent I wonder what educational impact that the technology will have, especially literature. I suppose that teachers could have essays written in class, where they would have environmental control. Or to allow AI assistance and then have subject matter discussion and testing based on the essay submitted.

2 Likes
11

It’s all in the framing. If she hears that after waiting for weeks, you could finally talk to an AI that was rumored to have huge potential and that would possibly kill jobs and entire sectors and change the world, and then your first question was asking it for romantic date ideas with your wife

Well, let’s say that if I were your wife, my heart would melt at that.

5 Likes
12

Cannot say but have seen some comments that are along the lines of that its use would be encouraged - like the calculator was.

I need to spend more time using it and do more than get it to write crappy short stories.

3 Likes
13

„Please write a mudspike forums poem“

In the realm of Mudspike, where aviators dwell,
A vibrant community, with stories to tell.
The forums alight with passion and cheer,
A sanctuary for pilots, a haven sincere.

From virtual skies to the digital seas,
Our members explore with grace and ease.
We share our knowledge, experiences, and lore,
Together we conquer, and always seek more.

In the threads of the forums, our journey begins,
With tales of adventures, losses, and wins.
From the roar of the engines to the hum of the blades,
We forge everlasting friendships and escapades.

In the cockpit we sit, hands steady and true,
Our eyes on the horizon, skies sapphire blue.
The Mudspike family, a bond we all share,
A love for the flight, a passion we wear.

We challenge each other, through trials and strife,
In the virtual battles, we sharpen our life.
Through the haze of the clouds, a beacon we see,
The Mudspike forums, where we all long to be.

So raise up your throttles and tighten your straps,
For the Mudspike forums are where it all happens.
From the dawn’s first light to the dusk’s last ember,
In Mudspike, we gather, and forever remember.

8 Likes
14

As an AI language model, I’m unable to create software directly. However, I can help you outline a plan for creating a fully dynamic campaign engine for DCS (Digital Combat Simulator), similar to Falcon 4 BMS. Here are the main components you’ll need to focus on:

  1. Dynamic mission generator:
    Create a system that generates missions based on the current state of the campaign, considering factors like frontline positions, available resources, and overall strategic objectives.

  2. Realistic AI behavior:
    Develop AI behavior for both friendly and enemy forces that simulates realistic tactics, decision-making, and communication. This will help create an immersive and challenging experience for the player.

  3. Persistent world state:
    Implement a system that keeps track of the state of the campaign, including unit positions, resources, and events. This will ensure that player actions have lasting consequences and create a sense of continuity.

  4. Scalable conflict:
    Design the campaign engine to scale the intensity and complexity of the conflict based on player actions and the overall state of the campaign. This will provide a dynamic and engaging experience throughout the campaign.

  5. Logistics and supply system:
    Integrate a logistics and supply system that simulates the allocation and transportation of resources, such as fuel, ammunition, and reinforcements. This will add depth to the campaign and force players to consider the strategic implications of their actions.

  6. Weather and time system:
    Incorporate a realistic weather and time system that affects mission planning, unit performance, and visibility. This will add an extra layer of immersion and challenge to the campaign.

  7. Briefing and debriefing system:
    Create a system that generates mission briefings and debriefings based on the state of the campaign and player actions. This will help set the context and provide feedback on the player’s performance.

  8. Customizable difficulty settings:
    Allow players to customize the difficulty of the campaign by adjusting factors like enemy skill, resource availability, and mission complexity.

To develop this campaign engine, you’ll need a team of programmers, game designers, and testers with experience in game development, AI programming, and simulation systems. You’ll also need to collaborate with the DCS community and Eagle Dynamics (the developer of DCS) to ensure compatibility and integration with the base game.

5 Likes
15

Please tell me you wrote that.

4 Likes
16

Straight copy and paste out of GPT4

6 Likes
17

You know, I’m but a simple gardener, but that thing effin’ scares me. It looks like a duck, walks like a duck and it quacks like a duck. But yeah sure it’s just a glorified word completer. Until it decides the trains run far more efficiently if it shoves all of us in the gas chamber.

11 Likes
18

Yeah, why does it feel like being the moth flying into the flame?

They have a safety/morale standard, well at least they try:
Usage policies (openai.com).

Imagine how many of these machines will be built without. Just a wide open abyss staring into you.

Right now I’m still scared more by humans than by computers. Let’s see when this tips over. OpenAI and Boston Dynamics merger, anyone?

2 Likes
19

Coincidentally there was a scaremongering AGI article in the Murdoch press here yesterday.

They completely forgot to mention that.

Like all AGI/AI, it can only do what it has been trained to do
 for now (even if on occaison we can’t figure out why it gave the answer it did)!

Absolutely. It won’t be the end of human creativity. Sure some students will use it. There has always been those who will always take the lazy/easy way out, but not all
 It might make ‘human certified’ art more valuable?

I also hesitate to call it cheating, because they are learning to use a tool that they would use outside of an educational institution anyway?

I remember a maths teacher telling us that we needed to learn how to do it the hard way because we won’t be carrying a calculator with us wherever we go
 ‘cough’ Smart Phone ‘cough’.

3 Likes
20

Negative, Ghost Rider.

2 Likes