ChatGPT and the like are neither artificial nor intelligent, it's garbage in garbarge out model that reckons Israelis deserve to have their own state like any other free people but Palestinians don't (literally what it responded in response to question on Israel/Palestinian conflict).
It's a massive garbage dump of all web-based data models and inherent biases of the programmers/developers. Neither intelligent nor artificial.
Well, ChatGPT is revolutionaries is because it has an association thinking function on AI. And that is one of the backbone and one of the major issue for AI development.
Machine don't think like we human do. Machine process data whenever we put it, so if I put in an instruction saying 1+1 = 2 on a machine, that instruction is ALWAYS going to give you a "True" answer. Becuase that is how Machine work.
However, people uses association to their memory. Which mean we don't generally process information the same way a Machine does, for example, if I ask you, what did do in the afternoon 209 days ago, most of the people will not be able to remember using this association, unless you have a photographic memory, but if I say what did you do in the afternoon of the last day of Ramadan, you probably will remember given if it wasn't too far ago.
Problem is, Machine learning does not really do this because that, because as you said machine process information garbage in, and garbage out. You need to either absorb all the information on all of what you did and then build a database around it, and then using it to retrieve the information you need. Or have an part of an associative function without the user input. But then the association part is still an issue, because yes, you can build hundred if not thousand or millions of database to store all the world information (Technically it's possible but I don't think even Petabyte data can cut it). Which make this ChatGPT query is unique
We all know it's Maggie who shots Mr Burns, so the answer is not really that unique and special, the special part is, how did ChatGPT knows I am asking about Simpsons. Notice that this is an open question implying a Simpson character, it did not give me a list of Mr Burn that being shot in all of its database, which mean it would have to somehow "Know" and more importantly "Chooses" the Mr Burns I refer to is the one in the Simpsons, and not this guy Jaylen Burns that was shot last month
An affidavit alleges Brown shot Burns with a handgun at an apartment complex on Jackson State’s campus. The judge in […] The post Suspect arrested in death of Jaylen Burns, who was shot on Jackson State campus appeared first on TheGrio.
news.yahoo.com
The process of eliminate all Burns and giving the one that I imply, but I did not put in specific instruction for it to look for Simpsons is all but breaking the usual instruction protocol. given that it was not hard coded in its system,
Again, unless this is hardcoded in its core code (which mean the dev anticipated someone is going to ask the Simpson question) if this is actually machine learning. then this is a quantum leap because you are allowing the machine to think outside its usual instruction protocol, if this is unlocked, then it means you have a machine which thinks and process information exponentially faster than any of us could, and that's scary, because that open to all sort of moral and righteous decision making process, Such as this
Or if it goes further, then we can literally ask ChatGPT to lie (Which it can't at the moment), if that happen, then ChatGPT can do what 95% of what a Human can do in term of thought process.
Industrial process won't be able to replace human, because it's one set of parameters that's always happen in an industrial setting, eg, how to make things faster, how to make cost lower, etc those are all very common question for AI to slimline, which mean you don't really need good AI to do that optimization, what you need is a giant calculator that process everything for you. But ChatGPT or other Chatbot alike is changing the way a computer "think" that is more important regarding AI development. I mean how would you beat me if my AI can think roughly 100000 times of a human counterpart, what that person can do, my AI can do that too and it will do it 100,000 times faster and all its interaction when your AI is just good at optimization??
www.lifewire.com
