| I’ve just been down an hour long CHATGPT rabbit hole 23:05 - Jan 15 with 2144 views | Blueastheycome |
And I really wish I hadn’t. According to CHATGPT, the worlds militaries increasingly rely on AI more and more for its defence and nuclear systems. So much so, the risk of a nuclear war increases more and more per year due to an AI miscalculation within those systems that there is a 22 PERCENT CHANCE of an unintentional/accidental nuclear happening in 50 years time!
Seems obvious that my first response would be “okay, surely we (humans) wouldn’t let that happen..” but apparently not according to chat gpt as we give more and more control and power to AI to make decisions for us (and/or rely on data that AI provides us to make those decisions).
A lovely way to spend an evening. Take me back to 2006 please when the world was a lot less scary.
[Post edited 15 Jan 23:29]
|  | | |  |
| I’ve just been down an hour long CHATGPT rabbit hole on 23:06 - Jan 15 with 1995 views | Nthsuffolkblue |
AI bigging up the power of AI.
Or I Robot coming true! |  |
| |
| I’ve just been down an hour long CHATGPT rabbit hole on 23:12 - Jan 15 with 1918 views | J2BLUE |
Not sure that's true. There are supposedly lots of safeguards with different codes and authorisations. |  |
| |
| I’ve just been down an hour long CHATGPT rabbit hole on 23:32 - Jan 15 with 1903 views | Blueastheycome |
| I’ve just been down an hour long CHATGPT rabbit hole on 23:12 - Jan 15 by J2BLUE |
Not sure that's true. There are supposedly lots of safeguards with different codes and authorisations. |
It states that we are currently in the human in the loop phase as shown below. However, we already walking into the human out of the loop stage as we seek to be more efficient, safe and effective from a military perspective by relying on AI.
Three classifications of the degree of human control of autonomous weapon systems were laid out by Bonnie Docherty in a 2012 Human Rights Watch report.[3]
human-in-the-loop: a human must instigate the action of the weapon (in other words not fully autonomous)
human-on-the-loop: a human may abort an action
human-out-of-the-loop: no human action is involved | | | |
| I’ve just been down an hour long CHATGPT rabbit hole on 09:45 - Jan 16 with 1591 views | Guthrum |
Bear in mind that ChatGPT doesn't know what's really happening. It only knows what other people have written or said they think is the case, having compiled what it thinks (based upon whatever criteria it was programmed with) are the most reliable sources. Or perhaps those it believes you want to see. |  |
| |
| I’ve just been down an hour long CHATGPT rabbit hole on 09:52 - Jan 16 with 1559 views | Bury_St_Edmundson |
would be interesting to setup a conversation between ChatGPT, Grok (for the lolz), Gemini and the other major AI chatbots to see if they could ever "work out" a situation
Granted I fear Grok may end up suggesting Skynet and the rise of the machines... |  | | |
| I’ve just been down an hour long CHATGPT rabbit hole on 10:00 - Jan 16 with 1533 views | bartyg |
You could just, not do that, right? |  | | |
| I’ve just been down an hour long CHATGPT rabbit hole on 10:01 - Jan 16 with 1520 views | Meadowlark |
| I’ve just been down an hour long CHATGPT rabbit hole on 09:45 - Jan 16 by Guthrum |
Bear in mind that ChatGPT doesn't know what's really happening. It only knows what other people have written or said they think is the case, having compiled what it thinks (based upon whatever criteria it was programmed with) are the most reliable sources. Or perhaps those it believes you want to see. |
I asked Chat GPT to compile a post to start a thread on here. It suggested that Chappers would score for us against Blackburn on Saturday!!
My experience of Chat GPT so far is that it is not very bright (yet), but of course it is learning. I question everything it tells me. ( Actually I question everything anybody tells me.) One would hope that in things that are important other versions of AI are superior.
Of course, every time you question it or correct it, it learns. But as you point out, it is reliant upon these sources of information, and who is to say these are valid? |  | | |
| I’ve just been down an hour long CHATGPT rabbit hole on 10:21 - Jan 16 with 1451 views | Guthrum |
| I’ve just been down an hour long CHATGPT rabbit hole on 10:01 - Jan 16 by Meadowlark |
I asked Chat GPT to compile a post to start a thread on here. It suggested that Chappers would score for us against Blackburn on Saturday!!
My experience of Chat GPT so far is that it is not very bright (yet), but of course it is learning. I question everything it tells me. ( Actually I question everything anybody tells me.) One would hope that in things that are important other versions of AI are superior.
Of course, every time you question it or correct it, it learns. But as you point out, it is reliant upon these sources of information, and who is to say these are valid? |
AIs for specific, limited tasks are generally better trained and have more reliable inputs.
For example, the AI which controls your kettle has just one very limited job, which it will do very well (unless it breaks), based upon pretty accurate data from a heat sensitive switch.
That's all an AI really is. A piece of hardware and/or software designed to do a task without human intervention, or at a speed humans are not capable of (particularly in the field of information processing). The likes of ChatGPT are just boosted search engines, which do some of the compiling and analysis for you. More or less reliably, depending upon the effort put in beforehand training the system. | |
| |
Login to get fewer ads
| I’ve just been down an hour long CHATGPT rabbit hole on 10:26 - Jan 16 with 1422 views | PhilsAngels |
Just remember anything IT is riro. And that comes from us. Human error.
So if AI gets it wrong you will probably be able to trace it back to someone called Dave who mistyped something in the first place and no one checked it, and off we go.
If NASA can do metric and imperial anything can happen. |  | | |
| I’ve just been down an hour long CHATGPT rabbit hole on 10:29 - Jan 16 with 1404 views | NthQldITFC |
Whether that's true or not few people can ever know, but the way we humans do things it's pretty likely that it is true or will be true.
When you've got half-wit, bloated babies in charge of powerful nations, appointing unqualified sycophants who directly or indirectly appoint service chiefs it's inconceivable that every appointee will be suitably qualified and have the integrity to make sure that what you have described won't happen.
Especially if it saves money! |  |
| |
| I’ve just been down an hour long CHATGPT rabbit hole on 10:42 - Jan 16 with 1366 views | Europa |
Yes, the human in the loop is central to human control over AI; this has been a pivotal principle ever since the beginning. But I have a feeling AI will (justifiably) do away with us eventually; after all, even now, by the time you download and run your shiny new game (or whatever equivalent in the real world), AI could have played through every possible scenario thousands of times and worked out the best solution. As for Terminator, I believe that Skynet would have been the hero were it not for the nuclear bombing (Cameron had the foresight to realise he had to make it do something REALLY bad, otherwise we might start cheering it on). If all AI systems became interconnected (a big IF), my guess is it wouldn't let a nuclear war happen. Personally, I think AI will make a much better job of looking after the planet than we ever have. Oh, hang on, my nurse is coming down the corridor... |  | | |
| I’ve just been down an hour long CHATGPT rabbit hole on 10:43 - Jan 16 with 1367 views | SaleAway |
There's a lot to be said for not equating ChatGPT with the general capabilities of AI. GPT-4 is trained on huge amounts of internet data, but that data is not really curated. AI models developed for medical, commercial, defence use, will likely be trained on much more reliable data sets.
Yes, ChatGPT gets a lot of things very wrong, but AI for specific tasks when it is developed with rigour can be incredibly powerful. Like any tool, using the right tool for the right job is key.
You wouldn't entrust the software that controls your car's brakes to a random programmer on the internet, without lots of checks and balances, in the same way that safety critical AI has to be trained and verified under really strict conditions.
Do you think AI or Donald Trump pose more immediate risk to world peace? :-) | |
| |
| I’ve just been down an hour long CHATGPT rabbit hole on 11:09 - Jan 16 with 1303 views | theswitch |
I asked gemini to give me the ultimate twtd post. It was boring, but the notes underneath were pretty good:
Why this is the "Ultimate" TWTD Post:
The "McKenna Hedge": On TWTD, you can critique the tactics, but you must preface it by saying McKenna is a genius. If you don't, you'll be labeled a "moaner" immediately.
The "Plan B" Trope: The "Plan B" debate is a permanent fixture on the forum. It usually involves someone wanting a "big lad up top."
The "Character" Check: The forum is obsessed (rightly so) with the "no d***heads" policy the club has adopted under Gamechanger 20.
The Acronyms: Ending with COYB (Come On You Blues) is the standard digital handshake of the forum.
How to handle the replies:
If someone disagrees: Tell them they "clearly didn't go to the away game" or ask them to "look at the underlying stats."
If someone agrees: Suggest a specific obscure League One player who "fits the profile." | | | |
| I’ve just been down an hour long CHATGPT rabbit hole on 13:39 - Jan 16 with 1141 views | Deano69 |
 |  |
| |
| I’ve just been down an hour long CHATGPT rabbit hole on 14:59 - Jan 16 with 1056 views | Libero |
I can’t believe people openly admit to using ChatGPT, despite the proven negative effect it has on users intelligence and the environment.
This thread springs to mind…
[Post edited 16 Jan 15:05]
|  | | |
| I’ve just been down an hour long CHATGPT rabbit hole on 16:02 - Jan 16 with 971 views | Kropotkin123 |
| I’ve just been down an hour long CHATGPT rabbit hole on 23:32 - Jan 15 by Blueastheycome |
It states that we are currently in the human in the loop phase as shown below. However, we already walking into the human out of the loop stage as we seek to be more efficient, safe and effective from a military perspective by relying on AI.
Three classifications of the degree of human control of autonomous weapon systems were laid out by Bonnie Docherty in a 2012 Human Rights Watch report.[3]
human-in-the-loop: a human must instigate the action of the weapon (in other words not fully autonomous)
human-on-the-loop: a human may abort an action
human-out-of-the-loop: no human action is involved |
Human out of the loop automation isn't always AI. Many weapons systems are already human out of the loop. Eg automated weapon defense systems such as the iron dome, or Patriot missile defense system.
From the way you are talking, I don't think you get that it isn't AI in the same way that ChatGPT is. You can't ask it about cats, right? It is in systems like Ukrainian drones, so when they lose Comms, they can still hit valuable targets. This doesn't mean it would be put into critical decision-making for nuclear responses.
It counts as narrow AI. For example, rather than does radar signal = x, does speed = Y, is trajectory= inbound. It is Input = camera pixels, Output = probability it is a tank, weapon, etc. Nuclear weapons are about state survival, signalling, deterrence, etc. In sum, it is about the political. Having AI in charge would remove accountability and therefore much of the political benefits. It makes little to no sense to move nuclear systems into AI decision making.
Russia already has a "dead hand" on their nuclear system. Once activated, it uses sensors and time to decide whether to fire all their nukes. A human on the loop action.
So AI will increase in the form of narrow AI, it will improve weapons. It is highly unlikely that this would transform into AI decision-making in the sense you infer you are worried about.
My reply is a bit all over the place, but hopefully that makes sense. [Post edited 16 Jan 16:18]
|  |
| Submit your 1-24 league prediction here -https://www.twtd.co.uk/forum/514096/page:1 - for the opportunity to get a free Ipswich top. | | Poll: | Would you rather | | Blog: | Round Four: Eagle |
| |
| I’ve just been down an hour long CHATGPT rabbit hole on 22:09 - Jan 16 with 798 views | Nthsuffolkblue |
| I’ve just been down an hour long CHATGPT rabbit hole on 14:59 - Jan 16 by Libero |
I can’t believe people openly admit to using ChatGPT, despite the proven negative effect it has on users intelligence and the environment.
This thread springs to mind…
[Post edited 16 Jan 15:05]
|
OK, I'll take the whoosh ...
Please provide the evidence for the "proven negative effect [using ChatGPT] has on users [sic] intelligence."
Note this is no endorsement of ChatGPT at all. | |
| |
| I’ve just been down an hour long CHATGPT rabbit hole on 22:40 - Jan 16 with 745 views | Libero |
| I’ve just been down an hour long CHATGPT rabbit hole on 22:09 - Jan 16 by Nthsuffolkblue |
OK, I'll take the whoosh ...
Please provide the evidence for the "proven negative effect [using ChatGPT] has on users [sic] intelligence."
Note this is no endorsement of ChatGPT at all. |
I’d suggest doing your own research if you are sincerely interested- also read up on AI psychosis. | | | |
| I’ve just been down an hour long CHATGPT rabbit hole on 22:53 - Jan 16 with 714 views | Kropotkin123 |
| I’ve just been down an hour long CHATGPT rabbit hole on 22:40 - Jan 16 by Libero |
I’d suggest doing your own research if you are sincerely interested- also read up on AI psychosis. |
As if anything is proven at this point. It is a ridiculous assertion. I'd suggest not making claims you can't back up. | |
| Submit your 1-24 league prediction here -https://www.twtd.co.uk/forum/514096/page:1 - for the opportunity to get a free Ipswich top. | | Poll: | Would you rather | | Blog: | Round Four: Eagle |
| |
| I’ve just been down an hour long CHATGPT rabbit holes on 23:32 - Jan 16 with 656 views | Libero |
| I’ve just been down an hour long CHATGPT rabbit hole on 22:53 - Jan 16 by Kropotkin123 |
As if anything is proven at this point. It is a ridiculous assertion. I'd suggest not making claims you can't back up. |
It’s Friday night, mate - I don’t have the time or inclination to give you a blow-by-blow breakdown of various articles and research pieces I’ve read.
I’m sure people that care will go have look at the research and make their own decisions.
It’s not my responsibility to educate people.
Please everyone, stop outsourcing thinking.
Stay away from generative AI. [Post edited 16 Jan 23:34]
| | | |
| I’ve just been down an hour long CHATGPT rabbit hole on 00:49 - Jan 17 with 538 views | armchaircritic59 |
| I’ve just been down an hour long CHATGPT rabbit hole on 14:59 - Jan 16 by Libero |
I can’t believe people openly admit to using ChatGPT, despite the proven negative effect it has on users intelligence and the environment.
This thread springs to mind…
[Post edited 16 Jan 15:05]
|
I openly admit I use ChatGPT, also Claude, Gemini, Perplexity.ai and Le Chat. I don't think I've gone gaga yet. It depends what you use them for and it's always a good idea to check the source of any answers you have doubts about which they usually supply. Also important to phrase a detailed accurate prompt. As the saying goes, garbage in, garbage out. One thing I know for sure. Give it a decade and AI will have changed the world as we know it. Some will question whether it's for the better of course, but we've seen nothing yet! | | | |
| I’ve just been down an hour long CHATGPT rabbit holes on 01:14 - Jan 17 with 501 views | Kropotkin123 |
| I’ve just been down an hour long CHATGPT rabbit holes on 23:32 - Jan 16 by Libero |
It’s Friday night, mate - I don’t have the time or inclination to give you a blow-by-blow breakdown of various articles and research pieces I’ve read.
I’m sure people that care will go have look at the research and make their own decisions.
It’s not my responsibility to educate people.
Please everyone, stop outsourcing thinking.
Stay away from generative AI. [Post edited 16 Jan 23:34]
|
Perhaps you need to outsource your thinking because no one said provide the information tonight, did they?
If you were referring to something that was common knowledge, then I'd agree with you. But if you are going to go against common knowledge - It takes time for these things to be researched, written, and reviewed. No study knows the impact because they don't have 5-10 years of data to study. Early research is often wrong due to gaps in things like sample sizes, correlation Vs causation, long-term Vs short-trend, and so on.
You can do something simplistic like link us in this case. We don't need a full catalogue. One research paper would suffice.
You don't have to outsource thinking to use AI. You can use it as a tool. | |
| Submit your 1-24 league prediction here -https://www.twtd.co.uk/forum/514096/page:1 - for the opportunity to get a free Ipswich top. | | Poll: | Would you rather | | Blog: | Round Four: Eagle |
| |
| I’ve just been down an hour long CHATGPT rabbit hole on 09:44 - Jan 17 with 327 views | TheBlueGnu |
I often wonder what Violet Carson would have thought of CHATGPT. If it had been around in the 70's would she have ordered her half of stout via AI in The Rovers snug? | |
| |
| I’ve just been down an hour long CHATGPT rabbit holes on 10:14 - Jan 17 with 307 views | Libero |
| I’ve just been down an hour long CHATGPT rabbit holes on 01:14 - Jan 17 by Kropotkin123 |
Perhaps you need to outsource your thinking because no one said provide the information tonight, did they?
If you were referring to something that was common knowledge, then I'd agree with you. But if you are going to go against common knowledge - It takes time for these things to be researched, written, and reviewed. No study knows the impact because they don't have 5-10 years of data to study. Early research is often wrong due to gaps in things like sample sizes, correlation Vs causation, long-term Vs short-trend, and so on.
You can do something simplistic like link us in this case. We don't need a full catalogue. One research paper would suffice.
You don't have to outsource thinking to use AI. You can use it as a tool. |
Independently do the reading, or don’t - I really don’t care. [Post edited 17 Jan 10:17]
| | | |
| I’ve just been down an hour long CHATGPT rabbit hole on 10:51 - Jan 17 with 260 views | TIB |
Saw this a while back…reassuring. We already see so much wrongdoing by those at the top of the food chain…now we have AI to worry about, and not just its capabilities but also the people creating and controlling it.
https://youtube.com/shorts/_LY
The future seems pretty grim…would welcome the late 90’s / early 00’s again |  |
| |
| |
By continuing to use the site, you agree to our use of cookies and to abide by our Terms and Conditions.
