I Interviewed ChatGPT — And It Lied To Me

Forbes contributor Barry Collins joins ”Forbes Talks” to discuss Chat GPT.

Read the full story on Forbes:

Subscribe to FORBES:

Stay Connected
Forbes newsletters:
Forbes on Facebook:
Forbes Video on Twitter:
Forbes Video on Instagram:
More From Forbes:

Forbes covers the intersection of entrepreneurship, wealth, technology, business and lifestyle with a focus on people and success.

Welcome to Forbes talks joining me now Is Barry Collins a contributor here at Forbes Barry thank you so much for Joining me hi Great so I want to talk about chat GPT Which has really gained popularity Recently first can you tell us what it Is Well it’s it’s an AI system that Basically Starts to chat like a human almost scary Like a human in fact uh it’s the latest Version of this system that’s been Released by a company called open AI Which Elon Musk was involved with at the Start and uh as various investors so Yeah it’s a it’s a big field at the Moment and uh this is just one of the Latest developments and how recently did This latest development come out Uh only the last week or so it’s it’s Suddenly gone massive on social media I’m sure lots of people have seen Conversations being uh posted on Twitter And stuff but yeah it’s it’s a real real New uh arrival on the scene and why do You think it has gone so viral I think just because the sheer Sophistication of the conversations and The range of stuff it can do I I sort of Interviewed it it’s uh it’s right in Poetry it’s writing songs it’s writing Tweets headlines stories blog posts I Mean you can there’s pretty much

Anything you ask it to do it at least Has a stab at responding Um and it’s pretty terrifying some of The stuff it’s doing So you interviewed it for Forbes and you Called it a liar in your piece can you Explain Yeah well I I was asking it about the Ethics of AI and and what it felt about Uh where it was ethical for for the AI To sort of behave like a human human Being and it gave me some pretty Straightforward answers and some pretty Good answers Um and then I was sort of running a bit Dry so I asked it to write a poem for me About the ethics of AI which it did and It did it twice in fact it had a sort of Stab at a more humorous version and it’s Not going to win any awards of poetry But uh it had a go so what level of Self-awareness I guess does this Technology have I don’t know I mean I don’t think it’s Self-aware I think it’s it’s trying to Just to sort of enter conversations like A human being and does so very Convincingly a lot of the time Um but I think it’s just sort of Generating answers based on lots of text From training that is it’s taken in the Past I don’t think we’re quite at the Point where the uh the AI is completely Self-aware and you asked it this

Question but I want to ask you what you Think what are the examples of the good That AI has done I mean where do we start I mean there’s Lots of stuff where I think AI is a badly misused Towing in Fact I think for a lot of me machine Learning is just called Ai and it all Falls under this bucket but there’s all Kinds of good that it’s doing in terms Of uh manipulating data so you know even Through the covered uh experience we saw Uh where to AI was being used to sort of Wear to Target resources and and stuff Like that so it’s had huge impact in the Medical fields and you know in the Technology fields that we work in in Terms of uh looking at data sets and Being able to intelligently decide where To use something Um but you know I think AI gets a bad Rap in the Press when people like me a Lot of the time but um it actually does A lot of good you also asked it if it Was misused do you think it’s misused Yeah I mean and in fact it gave me a Pretty good answer about cases where It’s been misused you know in terms of Badly trained data and uh it being used Uh to to sort of For stereotyping for example of certain Races and uh genders so you know there’s There’s definitely cases of misuse in The past and those cases are good like

Pretty much all kinds of Technology And I think one of the most interesting Questions that you asked it was do you Think an AI being trained to respond Like a human is ethical what do you Think about that Uh I think as long as you know that it’s an AI behind the scenes I’m okay with it I Think where it starts to get shady and You know we see it all the time on the Internet nowadays if you it says come And speak to our advisor and you know Full well that’s not a human behind that That’s a computer responding to me and That’s what I don’t like I think we’re Computers or AI pretend to be human Beings about letting you know I think that’s unethical I don’t think That’s correct uh I think if you’re told Up front you know you’re speaking to our AI assistant and it responds in a Human-like manner I’ve got no issue with That and what did you learn from your Experiment and your entire interview [Laughter] I’ll learn It it terrified me to some extent it was Um it’s amazing the degree to the Sophistication that what this AI can do Now and the degree of the arms as it Gave and the there wasn’t a lot of Nonsense in there there wasn’t a lot of Stuff that would immediately trigger me

To think this is an AI and this I knew It was and it was battering out answers At a rate that no human could Um but the stuff I’ve seen this week From chat GPT is just mind-blowing the Honestly it’s terrifying how how good it Is that the range of stuff it can do uh Whether that’s writing code writing Poetry uh writing blog posts it’s it’s Astonishing it really is what’s the most Staggering thing you’ve seen from chat GPT this Uh I think the part where someone Managed to convince it to write code and Write what’s called a virtual machine so It it actually created a piece of Software itself and then ran that on Amazon servers and I think they had to Shut that down pretty quickly because They they realized this thing was Basically generating code all by itself But and you know that that was just Someone playing with it and Experimenting me what it can do and Suddenly these things off writing its Own software code that that in itself is Is pretty staggering and pretty Dangerous in that you seem to be kind of In the middle of terrified of chat GPT And it’s completely astonished by it so Overall is your response what other People are feeling as well what has been The response to this Yeah I I think that is I think people

Are generally staggered about how good This thing is and then going You know blimey this is this is Terrifying stuff this is got the Potential to be absolutely huge and if It’s not controlled properly uh then it Has some pretty big implications as well I mean people will use it this week for Example uh to to write effectively Viruses uh to create a code that would Launch an attack on a website uh you Know you could easily see how something Is intended to be used uh harmlessly Suddenly gets misused in a in a very Scary way so how do Well that basically boils down to the Creators and open AI the company behind This they’ve got to be careful about the Parameters of what they allow this Software to do and and as we’ve seen uh Throughout the week as they allow this System to be used by anyone for free and We’ve seen that some of the stuff that It was doing at the beginning of the Week is now it’s now being curtailed so They’re sort of learning as they’re Going on themselves they’re stopping the Uh chat GPT from answering certain Questions or writing code in a certain Way Um as people have found these exploits If you like so are they giving Chad gbt In a sense a conscious I I think that’s how you were playing it

I don’t think it’s I don’t think it has A conscience Um I think it it it works within Parameters of what it can do and uh you Know as the AI said to it to me itself It it can only do what it’s it’s human Overlords if you like allow it to do uh It’s it’s to some extent it doesn’t have Free will it can only exist between Uh you know the A to Z of what what the The company allows it to do and I’m Seeing a lot of people like yourself Using this for fun on social media it’s Really been fun to see how it’s been Used but what do you see is the future For this software That that’s really hard to say I think It’s you know it could it could have so Many uses Um from from writing journalistic copy The stuff like I do uh to uh you know be In a chatbot on shopping Services I can Easily see this software Um fitting straight into that market Um that’s all kinds of things it could Do and all kinds of purposes it could be Put to Um and you know right now we’re just Seeing it in in text form of course if You marry that with voice and and Google’s already here to some extent and You know devices like Alexa but if you Can marry this kind of intelligent Conversation in AI we have a voice

Product as well You know the the implications there are Pretty huge the possibilities certainly Sound endless do you see in the future This technology replacing human Different jobs I think it really is just in some extent You know in in our industry financial Journalism there’s a certain amount of Financial journalism that’s written uh By AI already uh you know in all kinds Of industries that there’s AI nibbling At the edges if if not taken over human Jobs I think it’s it’s certainly Inevitable for a certain amount of For one of a better word drug work where It’s it’s just churning out copy uh I Think that that is very likely to to Move over to AI Um and I think the more creative stuff Uh is is more likely to remain uh in Human hands hopefully fingers crossed we Will definitely keep our eye on this Barry thank you so Please thank you

You May Also Like

Leave a Reply

Your email address will not be published. Required fields are marked *