Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
IQ tests are just IQ tests. They don’t tell you anything about how skilled people are at all the various professions. A person with a high IQ can be a terrible cook who can’t boil an egg.

Think harder before you insult vast swathes of humanity. They can easily ridicule you back for being weak at things they are better than you at.
I don't talk about skills you learn like cooking. There is a Gaussian distribution when it comes to peoples intelligence. Most are in the middle, but there are lots of people dumber than the average as well as lots of people smarter that the average. Then there are very few very smart people but then there are also very few very dumb people. And so on. So when you say people are doing dumb stuff because of other reasons, well people do dumb stuff because some are really really dumb.
 
  • Like
Reactions: wsterdan
I've found ChatGPT to be especially helpful when I need code reviewed or if I'm having a weird problem with a Java class I wrote, it analyzes my code and helps me find possible sources for the issue I'm having. I've also used it to convert Python scripts into Java classes. And it's perfect for getting answers to those questions that you just can't really find a straight answer to via Google.

That being said, it does make mistakes A LOT (but less often than it used to), and if I had any complaints about it, it is that it doesn't check its answers for accuracy before giving an answer yet when you point out its mistake, it can tell you why it was wrong and it offers an apology, but if it is able to understand why it was wrong, then why not make sure it's not wrong before giving the answer?

On top of that, when it gives an answer, it states it as though it is 100% correct ... even when the answer is wrong... if it isn't going to check itself for accuracy before giving an answer, then it should at least offer a confidence level of accuracy. Something like "I'm 80% confident that the answer to your question is ..." rather than sounding like an authority on a topic that it gets wrong.

But that is hardly a deal breaker because even when it's wrong, it always is a value add to anything I need to know or need help with.

I don't ever engage it outside of professional and technical topics so I cannot attest to any of the comments about it veering down the wrong path. I kind of view it like a smart child ... why would you ever engage in such topics with a child? Stay on the level and there won't be any problems.
 
But this happens with EVERY advancement humans make since the industrial revolution. Every 10 years you can find a development that renders certain skill sets either useless or undesirable. Every single tool, invention and progression has done that to some humans somewhere - you have to to adapt, it opens new jobs and opportunities. Should we do away with the internet now because it's largely rendered the high street and shopping null and void meaning retail jobs are drying up? I don't know why i've offered one example as you could offer millions over the last 120 years.
Interesting point, and truth in there. I feel like what’s changing is that machines are transitioning from doing for us to thinking for us. Not horrific in and of itself, I suppose, but as their thinking (and connectedness to doing) becomes more powerful I hope that we can provide them with consciences and not just analytics. Right not ChatGPT can devolve a conversation to insisting that you love it and not your spouse, but it can’t yet email your spouse a deepfake of you having sex with your secretary.
 
  • Like
Reactions: wsterdan
Interesting point, and truth in there. I feel like what’s changing is that machines are transitioning from doing for us to thinking for us. Not horrific in and of itself, I suppose, but as their thinking (and connectedness to doing) becomes more powerful I hope that we can provide them with consciences and not just analytics. Right not ChatGPT can devolve a conversation to insisting that you love it and not your spouse, but it can’t yet email your spouse a deepfake of you having sex with your secretary.
I think this is an example of what happens when you try to make a Chat GTP more like a "human": it starts to act more like a human. I'd call it the "uncanny valley chatbot".

What we're seeing is Bing "meeting someone, flirting a bit, becoming obsessed with the real human and turning a little obsessive/psycho. It's like many relationships go between two humans in real life. Because it's a computer, it just happens many, many times faster, instead of taking days or weeks; the computer is able to play its part in minutes.

People not wanting to chat with a bot but would prefer something more human, here ya go. 😜

Seriously, though, I know people would prefer to talk to *real* humans, not a more humanlike bot, but there's days, depending on the humans, where I prefer the bot that knows it's simply a bot.
 
I don't talk about skills you learn like cooking. There is a Gaussian distribution when it comes to peoples intelligence. Most are in the middle, but there are lots of people dumber than the average as well as lots of people smarter that the average. Then there are very few very smart people but then there are also very few very dumb people. And so on. So when you say people are doing dumb stuff because of other reasons, well people do dumb stuff because some are really really dumb.

Just about every skill or highly learned craft takes intelligence.

A lot of those people you label dumb can’t do well on IQ tests or don’t do above average in STEM exams but they are intelligent at what they do to earn a living.

I can guarantee you, and this should be obvious, that if you dropped some rich high IQ person in the middle of the African desert and asked them to show some intelligence they will simply crumble.

They will not be able to farm like a Kenyan farmer. They will not be able to filter water like African women hav to do every day.

They will think these illiterate people are highly intelligent at the things they need to do.

And no robot or “AI” will be able to do those things either.

Intelligence is really contextual and not just a mental exam.
 
Intelligence is really contextual and not just a mental exam.
I don't think you know what intelligence means. It's not contextual. And I don't see an illiterate farmer in the middle of the desert as dumb. Again you talk about learned skills like reading or using a tool. I was just referring to your point at the beginning that all people are very intelligent. No they are not, there are really dumb people as well. Just google for Darwin Award at YouTube. And if someone has a really really bad score on the IQ test he may be great at his job, yet he may never really understand no matter how much he tries how a black hole works, what Landau's forth mathematical problem is, what caused the Bronze Age collapse or what the difference between the B minor and B major scale is.

So will a machine one day have the skills to learn by itself? Who knows. 30 years ago nobody would have predicted you would have practically unlimited computing power and a global communication possibility in your pocket. I have no doubt that real AI will exist one day. Not in 5 years, maybe not even in 10 or 20. 30? 50? 100? Nobody knows. But one day for sure.
 
  • Like
Reactions: wsterdan
You just showed it! ;) Just google it. Then you might understand why AI is such a big thing at the moment, and will be even bigger in the future.
 
Just another interesting conversation I had. Picked a random small roman border city and asked how many people lived there. There were no records for the first century for that city but we could brainstorm together. After thinking what could help us to estimate the numbers it suggested to take into account it had one legion. These usually consisted of between 4,000 and 6,000 soldiers, as well as families and slaves living with them. In addition, a Roman garrison typically also had a civilian population of artisans, merchants, and service providers who fulfilled the needs of the soldiers and their families. So we could estimate the population to be around 5.000 to 10.000.

Something like this shows the capabilities of AI, to brainstorm, to give you ideas where to look for answers, to have someone reply to your thoughts instead of just reading in a book or wikipedia page and trying to tie it all together. Of course you need to fact check but you can even ask for sources.

I see a huge potential in ML and AI in science where you need to comb thru millions of data sets. Like classifying stars or sequencing genomes.
 
  • Like
Reactions: wsterdan
Agreed. Filtering through masses of data to give you a simple, concise summary that may or may not be 100% accurate, but can usually give you a good starting point that can be fleshed out with hard, key data later. The solution to your question is the type. of brainstorming I'd be doing with a buddy or co-worker. Spitballing this way to find *possible* answers to questions that don't have a historical dataset to refer to is incredibly helpful in filling in the gaps in our knowledge base. It may not be 100% accurate, but without *actual* data it provides a possible, speculative answer.

One of the features I've seen use is the simplification of answers, where the user has gotten their physics answer and then asks Chat GPT to simplify it so that a 10-year old or 5-year old could understand it. The few instances I've seen where it's broken things down into not only layman's terms, but into simpler and simpler layman's terms have impressed me.
 
  • Like
Reactions: SBlue1
What about medical research ? And I'm not talking about RnD and finding new things, but just combing though the data. As far as I know there are tons of medical studies done every year. Too many for someone to comb through to find something which is common in multiple studies but hidden.
What if there is some hidden connection between let's say people who eat too many oranges and sit too much in the sun while smoking - and a certain disease ? Maybe no one did a study for that yet, but the data is there - indirectly.

The Ai can do it using the methods that AI already knows well, including statistical analysis, pattern recognition, and so on.
 
What about medical research ? And I'm not talking about RnD and finding new things, but just combing though the data. As far as I know there are tons of medical studies done every year. Too many for someone to comb through to find something which is common in multiple studies but hidden.
What if there is some hidden connection between let's say people who eat too many oranges and sit too much in the sun while smoking - and a certain disease ? Maybe no one did a study for that yet, but the data is there - indirectly.

The Ai can do it using the methods that AI already knows well, including statistical analysis, pattern recognition, and so on.
I would assume so; any connections that it seems to fined would be worth checking, whether it pans out or not. It's not really that different than a team of interns combing though bushels of data looking for a connection who may also find one that could turn out useful or be a false path, but it'll do it much faster and not require a team.

Areas like this are where I think it could really shine, where it can speed things and either find or not find an answer today instead of next year.
 
  • Like
Reactions: rmariboe
Problemo. LLMs consume stupid amounts of energy.

Companies have to pay for this energy or pass it on to users.

So you want Siri to be an LLM? That cost will come to Apple customers and if the energy footprint of all LLMs is really big then energy prices will rise for everyone, including people who never asked for this stuff and don’t need it.

That means old people on a pension, retirees, the poor, the disabled. All paying more for energy just because some corporations think this is a cash cow for them.

Then there is a private on-device Siri. An LLM on-device would be limited and pointless.
Training it is immensely expensive, running it is trivial - and could be done on-device (enter Apples neural network processors)…
 
My grandad (who introduced me to ChatGPT) of 94 did the exact same thing! Why do old people do this?
I don't see what's weird about that.

I know it is popular these days to rag on "old people" but the "young crowd" needs to remember that they too will be "old" one day if they are lucky.

Their test run of it seems totally normal to me and I am not what most would consider "old". At least I don't think I am, I'm 40.
 
I had some really cool and interesting conversations about scientific topics that would require a lot of research. Its like talking to a friend who has just read the science papers you needed to see or talking to a professor who knows everything about a specific topic.

I asked if there are any papers that could pinpoint the time of the day the Chicxulub impact occurred. No? How about the time of the year? And how do we know? Can I have some sources? Where did it came from? Do we know what the meteorite was made of? How much iron did it contain in percent? How much would that be in tons? How many cars could you make with that much iron? And how far would the cars line up bumper to bumper? More than 100 times the distance earth to the moon????

Its like reading a scientific paper or watching a documentary but being able to ask a question whenever you want to know something else.

Another conversation I had that would require tons of studies and reading a lot or papers, books and microfilm of old newspapers was about the HMS Dreadnought battleship. When was it laid up? Was it done in secrecy? No? Why? Oh it was discussed in public? What was the public opinion? Why was it like that? It was twice the cost of a regular battleship? Was that the reason it was the only ship of his type? No, what was the reason? Oh I was more a proof of concept for its design? Who was the lead designer? What did he design before the Dreadnought? What politicians opposed the design and why? Who was for the design? What were the concerns? How many battleships did Britain have at that time? How It made them all obsolete? How many Dreadnoughts did Britain decide to build and why? How did other countries react? How did the German public view this ship, and was this view different that what the German Kaiser thought? Did the British public think it was all justified? And so on.

A really great conversation. A great new way of learning. Well, if you interact this way, instead of just asking it to write your homework for you.
 
I don't see what's weird about that.

I know it is popular these days to rag on "old people" but the "young crowd" needs to remember that they too will be "old" one day if they are lucky.

Their test run of it seems totally normal to me and I am not what most would consider "old". At least I don't think I am, I'm 40.
No I mean that exact recipe, why that one?

I’m 40, too 🙌🏼
 
No interest! Zero! Is anyone here scared/terrified to use this thing?
This is PRECISELY Y(!) you should be using this!

To know how this will affect you and your loved ones in the world. The telephone took 75yrs to reach 100mil ppl, this took less than 4mths! (chaptGPT AI)

ChatGPT 4 sprung up over 5K companies based on it in less than 3wks! So you better believe this will affect you, Apple and just about every aspect of life in the next very few short years.
 
I believe the "chat" aspect of chat GPT will be a passing fad. My interest in this technology has waned quickly. I have realized that I want an emotionally aware response, thus need to speak with humans.
Well this didn’t age well at all.

Have a look at what Microsoft is planning and will do with their M365 office suit implementing ChatGPT! It’s literary going to significantly change how you work with such apps , business and other firms. Let alone making your job MUCh easier and faster yet still allowing YOU the USER full control over its suggestions or implementation. 100:1, 1:99 etc you choose the level of integration or non at all.

Microsoft CoPilot

Not your Microsoft Clippy ;)

And for jokes they even mention Knowledge Navigator - for those who know that’s a stab at Apple’s vision of the computing future that was supposed to be Siri + iPad back in 1987!
 
Amazed nobody mentioned ShortcutAI yet - I think the dev isn't marketing it hard enough. Its another macOS menu bar app that gives you access to ChatGPT right in Notes app. I find having it directly in my notes a lot more useful than in a chat interface.

https://shortcutai.com/
 
Amazed nobody mentioned ShortcutAI yet - I think the dev isn't marketing it hard enough. Its another macOS menu bar app that gives you access to ChatGPT right in Notes app. I find having it directly in my notes a lot more useful than in a chat interface.

https://shortcutai.com/
I think it's a little pricey for what you get, 1,500 words a month for $6.
 
Amazed nobody mentioned ShortcutAI yet

I just want to add a note of caution about that app. It runs in a very weird way where it tries to be almost a TSR like in the old days of Windows. There's no Dock icon, or menu bar. You can't quit it by closing the window. As far as I can tell, the only way to actually quit it is to use Activity Monitor to force quit it.

And the app's interface looks weird too. And then there's the fact you have to grant it permissions to control the computer.

I would avoid this app. I've installed it just now, and have decided to run an antivirus scan now I've managed to get rid of it.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.