ChatGPT gets basic maths wrong. It can be racist and sexist. It can be seemingly very confident and authoritative, yet provide incorrect information. ChatGPT has no nostalgia and does not reflect on its past. This sounds quite human to me.
I was going to try it out the other day but it wanted me to create a login so was going to investigate first…
Sounds very human from that…
Everyone wants a login these days, don’t they!?!
Just because it can mimic a human doesn’t mean it is… many birds mimic sounds for example.
I am not suggesting that it is human. However, I feel that a good way to think of it is as the equivalent of a very knowledgable person who, like all of us, is not always right. This adviser can still be of great value in conceptual work, developing code, solving problems, researching issues, evaluating arguments, etc., but all the advice must be checked before it is relied upon.
I believe that the critics of ChatGPT and Google’s Bard misunderstand the purpose and utility of such tools.
Yes, I baulked at handing over my phone number to some startup.
Seeing more and more chatter about this ChatGPT… and exactly where it sits as far as AI is concerned… Is it the start of the curve or end…
There does seem to be a lot of hype, as discussed in:
However, people do seem to relate to the new Bing, ChatGPT etc. Is it just a better, more relatable way to seek information and help than Google searches. It can provide flawed results, but so does Google.
It surprised me to hear that it could create computer code. I don’t know what coding is in it, but - if it was able to basically just learn how to write code… go and teach itself… that’s kinda… OMG. OMG!
Sure it was failed code as noted in the article but - it was able to correct itself with some help. So… again as suggested - if this is the start of the curve… we soon have AI that can write code, get it wrong, then self learn how to get it right… and… that is sounding to me like - true AI.
Maybe not “self aware” AI, but something that can problem solve and learn and…
How does this turn out well for humankind? We’re the most destructive force in the universe… (Ok, maybe all sentient life is, not just us… well, though - dolphins and some apes are sentient and not as destructive… yet…)
Where was I… oh yes -
LOL Yes. Them too. I’m going to try watch BSG again - only got to about S2 last time round…
I told my 15yo about the above Youtuber RE ChatGPT writing computer code. This week, he admitted to me firstly that he has a ChatGPT account, and secondly that he tried getting it to write some code in Python for him, based on the YouTuber’s story. He said the first draft of script was very basic, but worked, and when he then told ChatGPT to use variables (or something or other) - the code improved drastically - “to a Year 10 level”. (My son is in Year 10…)
Not thrilled he didn’t ask about getting an account for ChatGPT (and… at my parenting skills, not knowing he had done this)… but thought it was interesting that he decided to try that out. I know he wouldn’t have tried passing it off as his own work (he loves programing at school). I guess looking at recent articles - being a programmer may not be a job in the future - rather, it will be - being an AI-wrangler…
It seems that ChatGPT and similar have now attained a basic competency in coding. More importantly, it can serve to conceptualise what is needed and provide a competent draft of the required code.
This ABC story provides new terms for the competency required to use these new AI tools that can generate stories, images, code, audio in the voice of famous people and more:
Yes - that was the article I’d seen too
My wife’s reaction to hearing that our son set up himself up with an account… was of course to check whether it’s considered ‘safe’… and apparently there’s concerns about kids using ChatGPT as it can essentially research using the whole of the web, and may come back with “inappropriate” answers. I believe the official version of the software has safeguards built in… though I have heard that people have found ways to “turn them off” and get an unfiltered experience…
I’m just still shocked that an AI can learn and write code… and then improve on that code with some further guidance…
Just watched a guy on YouTube (assuming was real) teach ChatGPT to write the code to make a replica Flappy Bird game. He gave it dot points of what the game was about, and with a number of revisions and improvements - created a playable game.
It is really just… Wow. Computer Programmer will not be a job in the future…
My current job - the first 1-3 hours of the day is matching up payments to customers. The system is already part automated IF the customer put the right code on the payment. Of course - many don’t cos they’re idiots… so then I have to manually search and figure out who owns the money. I have no doubt that ChatGPT could be taught to locate just as many customers as I can… and probably more.
There are seriously hundreds of thousands of jobs just in Australia that this kind of software can eradicate. I knew it was coming… but - Wow…
I have not even tried to get a chatGPT account, I admit to being curious, but not curious enough. Just finished watching Caprica (the BSG prequel for those not knowing) and it was my second watch and now I understand more of the issues in BSG. It probably doesn’t really relate to the current state of AI but is likely to, in the future. Kinda glad I wont be around to see it all come to fruition.
I grew up with killer robots in pop culture but also the knowledge that such tech was beyond us. Now though… it may not even come down to AI gaining a real magical sense of self awareness that is oft cited in fiction - just the ability of software to make some clever deductions… with enough access and ability to turn our own systems on us.
From what I’ve seen ChatGPT is currently a cross between a parrot - ask a question and it will seek info and then paraphrase what is finds… and a - ? - young kid? The fact it writes computer code, can be given an objective, “learn” to follow the correct syntax etc and produce a workable result - to me that’s more than just parroting…
The time frame is what’s going to be interesting… how quickly the tech advances and how long til it’s updating its own code….
And maybe I watched too many killer robots movies as a kid.
The easiest way to try out ChatGPT is to install Microsoft’s Bing browser. It includes a button to access a later version of ChatGPT.
I am not sure whether you need a Microsoft account, however I think opening a Microsoft account is free anyway.
Thanks for the info. I’ve had a microsoft account for years, so I might take a look, but its not top of my must do list.
This YT video is interesting
This is an interesting video also:
Elon Musk’s involvement with AI:
- One of the original investors and founders of the company that created ChatGPT in 2015
- Warns against further development of AI (17 April 23)
- Announces his own AI project TruthGPT (18 April 23)
What does he really think about AI?