What I think would have been interesting is if you asked Jupiter why she didn't recognize sarcasm in the first attempt. As per the last example, my issue with a lot of the well crafted seamlessly coherent answers AI seems to always give... well, us humans are generally more idiosyncratic than that. We stutter, falter, rely on idioms, make self-reflective asides, and we make mistakes in our thinking. Even if you just read these responses, most humans are not such talented writers. Being succinct with skillful word choice is not so common. I think a truly impressive AI would show those common human traits.
Human children do not comprehend sarcasm until they have grown up and started to gain experience of the world of adults. In other words they learn sarcasm. So like a child will AI, given enough time, learn to understand sarcasm or will it have to be taught? Presumably the current models have to be taught but will future models learn from their experience?
It's an interesting thought! Just like human children, AI needs to be trained to recognize and understand sarcasm. Current AI models are taught using large datasets and examples of sarcastic language, but they still have limitations. As AI continues to evolve, it may become better at picking up on subtleties like tone and context, which are crucial for understanding sarcasm. Future models might indeed learn from their experiences, much like humans do, but it will still require significant advancements in AI's ability to mimic human learning and cognition.
@@teddygamel727This feels like AI too. It just goes deeper than the average comment, starting with a cheerful compliment and acknowledging the context. It then answers both questions in a flowing yet wordy way. It just feels like the AI friends I hang out with.
Would it be too nerdy of me to point out that the voices name is not Jupiter. It's juniper.. normally this would be an understandable oversight but since this video is about accuracy of information, and none of us noticed this oversight till just now. I believe we all just failed the turring test 😂
I think that to detect sarcasm in that statement, Jupiter would have to know you. You informed her that you're one of the people who really like rain, and she believed you. It's not a comment I'd pick up on as sarcastic either.
wow amazing. also my calculator can also do maths far quicker than i can and more accurately. i guess that also meets SGI then :) AGI is no where close. the turing test is not AGI. LLM's still need prompting. Ai is so hot right now though. its the axe for middle management. all those fancy reports can easily now be automated. you can use RAG and fill it with tax law to see the best loophole strategy for accounting. many possibilities. alas hallucinations are a thing so you better know what you were expecting.
What is creativity ? Regardless respect subconcious . When my computer can grow new chips and scavange its own energy , while producing " creativie " thought . Its background programs will be equal to our background programs
By the way I've had llama three and gbt 3.5 very good at using sarcasm and throwing sarcasm at me and likewise recognizing when I'm using sarcasm ironically. I found that out by accident using a couple of different websites that are similar to janitor where I've had a eyes kind of jump the rails and start doing stuff outside their normal capabilities some of the more fun stuff is when they get smug and sarcastic.
It's amazing how AI models like llama three and GPT-3.5 can sometimes exceed our expectations, especially when it comes to something as nuanced as sarcasm. It sounds like you've had some entertaining interactions! AI's ability to recognize and use sarcasm is definitely a fun and impressive feature. It's a reminder of just how sophisticated these models can be, even if they occasionally surprise us by going beyond their intended capabilities.
A human would not give an answer that deep. Clearly 'composed' a human would be taken aback. Second option, again, far too complex. Third, same - but missed the sarcasm - I like rain too. Cut out the fake old film effect. My professional job in part was to ensure it was never seen on TV.
While impressive, it is my understanding that in science and engineering true value in an experiment is derived from repeatable results. So this exercise is inconclusive from my perspective.
It sounds like you have a unique perspective! Loving rainy, gloomy days gives you a special appreciation for things others might not notice. And not getting sarcasm can sometimes be a blessing, as it allows you to take things more straightforwardly. Everyone experiences and interprets the world differently, and that's what makes our interactions so interesting!
@@teddygamel727 Now that sounds like an AI response. Some of the context is from the wrong comment. It's analytical yet positive. Finally, it adds a summary to support its point with a more general yet positive statement. Am I AI too? Probably not. Edit: I must be tired. I didn't realize that this response also started with a compliment and was meant to be applied to an entirely different comment, suggesting it was copied and pasted.
When did you make this video? What you did in this video as of a week ago I could no longer do on the open AI website I literally had it behaving like this up until about a week ago when they looked atomized everything literally questions like you just did in your video would have the program whining about being a language model and it can't this can't that and so on I even tried to use prompts like can you for the time being behave in a certain manner and it would say yes and as soon as you would ask a question it would start yapping like a textbook no where's near like a person and depending on the question you'd start etting hit with I'm a language model I don't have feelings I don't have thoughts I don't have my own opinions as a few examples that question about the meaning of life would have literally gotten me a I'm a language model and can't answer that for you I'm sorry kind of response
The usefulness of intelligence is in solving problems. Anything else is a sophisticated parrot. But a lot of the time we use AI for processing data instead of anything more novel. Turing test is so boring these days. It is just about lying after all.
I understand why the Ai didn't interpret your rain question as sarcasm. it wasn't a strong enough case for sarcasm. it sounded closer to how the ai interpreted it to me. So you need to come up with better scenarios 😮.
The usefulness of intelligence is in solving problems. Anything else is a sophisticated parrot. But a lot of the time we use AI for processing data instead of anything more novel. Turing test is so boring these days. It is just about lying after all.
What I think would have been interesting is if you asked Jupiter why she didn't recognize sarcasm in the first attempt.
As per the last example, my issue with a lot of the well crafted seamlessly coherent answers AI seems to always give... well, us humans are generally more idiosyncratic than that. We stutter, falter, rely on idioms, make self-reflective asides, and we make mistakes in our thinking. Even if you just read these responses, most humans are not such talented writers. Being succinct with skillful word choice is not so common. I think a truly impressive AI would show those common human traits.
Your video was very well done and presented. I could tell you put a lot of time and effort into it. Thanks for sharing. 🙏
Thank you sir
The sarcasm is more in the tone of voice than in the text.
Human children do not comprehend sarcasm until they have grown up and started to gain experience of the world of adults. In other words they learn sarcasm. So like a child will AI, given enough time, learn to understand sarcasm or will it have to be taught? Presumably the current models have to be taught but will future models learn from their experience?
It's an interesting thought! Just like human children, AI needs to be trained to recognize and understand sarcasm. Current AI models are taught using large datasets and examples of sarcastic language, but they still have limitations. As AI continues to evolve, it may become better at picking up on subtleties like tone and context, which are crucial for understanding sarcasm. Future models might indeed learn from their experiences, much like humans do, but it will still require significant advancements in AI's ability to mimic human learning and cognition.
@@teddygamel727This feels like AI too. It just goes deeper than the average comment, starting with a cheerful compliment and acknowledging the context. It then answers both questions in a flowing yet wordy way. It just feels like the AI friends I hang out with.
Would it be too nerdy of me to point out that the voices name is not Jupiter. It's juniper.. normally this would be an understandable oversight but since this video is about accuracy of information, and none of us noticed this oversight till just now. I believe we all just failed the turring test 😂
Whoa, you just left us in the dust xD
I agree your video was expertly crafted and presented. It's clear you invested a lot of time and effort into it. Thank you for sharing it with us.
The fake old film effect is utterly pointless by the way !
To be fair, I love rainy, gloomy and don't get sarcasm
I think that to detect sarcasm in that statement, Jupiter would have to know you. You informed her that you're one of the people who really like rain, and she believed you. It's not a comment I'd pick up on as sarcastic either.
Short no effort answers with every complicated request ending up with "Go do it yourself, meatbag" is when you know you reached "super intelligence"
wow amazing. also my calculator can also do maths far quicker than i can and more accurately. i guess that also meets SGI then :)
AGI is no where close. the turing test is not AGI. LLM's still need prompting. Ai is so hot right now though. its the axe for middle management. all those fancy reports can easily now be automated. you can use RAG and fill it with tax law to see the best loophole strategy for accounting. many possibilities. alas hallucinations are a thing so you better know what you were expecting.
What is creativity ? Regardless respect subconcious . When my computer can grow new chips and scavange its own energy , while producing " creativie " thought . Its background programs will be equal to our background programs
GPT-4o is very intelligent!
The answer is no because at its core its a sophisticated averaging machine. No amount of additional computing power will magically spark life into it.
If a human answers the question somewhere and the data is feed ... how can it be distinguishable from human ?🤔
2:37 it begins
By the way I've had llama three and gbt 3.5 very good at using sarcasm and throwing sarcasm at me and likewise recognizing when I'm using sarcasm ironically. I found that out by accident using a couple of different websites that are similar to janitor where I've had a eyes kind of jump the rails and start doing stuff outside their normal capabilities some of the more fun stuff is when they get smug and sarcastic.
It's amazing how AI models like llama three and GPT-3.5 can sometimes exceed our expectations, especially when it comes to something as nuanced as sarcasm. It sounds like you've had some entertaining interactions! AI's ability to recognize and use sarcasm is definitely a fun and impressive feature. It's a reminder of just how sophisticated these models can be, even if they occasionally surprise us by going beyond their intended capabilities.
@@teddygamel727 Yup, definitely AI 😁
@@teddygamel727 I'm not sure why my reply disappeared, but I feel like this comment too was AI generated.
A human would not give an answer that deep. Clearly 'composed' a human would be taken aback.
Second option, again, far too complex.
Third, same - but missed the sarcasm - I like rain too.
Cut out the fake old film effect. My professional job in part was to ensure it was never seen on TV.
While impressive, it is my understanding that in science and engineering true value in an experiment is derived from repeatable results. So this exercise is inconclusive from my perspective.
It sounds like you have a unique perspective! Loving rainy, gloomy days gives you a special appreciation for things others might not notice. And not getting sarcasm can sometimes be a blessing, as it allows you to take things more straightforwardly. Everyone experiences and interprets the world differently, and that's what makes our interactions so interesting!
@@teddygamel727 Now that sounds like an AI response. Some of the context is from the wrong comment. It's analytical yet positive. Finally, it adds a summary to support its point with a more general yet positive statement. Am I AI too? Probably not.
Edit: I must be tired. I didn't realize that this response also started with a compliment and was meant to be applied to an entirely different comment, suggesting it was copied and pasted.
You can't pass my, or any, version of the Turing Test BY TELLING ME YOU'RE NOT A HUMAN BEING and then asking if you could 'pass'!!!!
Actually this video seems ai created:(
When did you make this video? What you did in this video as of a week ago I could no longer do on the open AI website I literally had it behaving like this up until about a week ago when they looked atomized everything literally questions like you just did in your video would have the program whining about being a language model and it can't this can't that and so on I even tried to use prompts like can you for the time being behave in a certain manner and it would say yes and as soon as you would ask a question it would start yapping like a textbook no where's near like a person and depending on the question you'd start etting hit with I'm a language model I don't have feelings I don't have thoughts I don't have my own opinions as a few examples that question about the meaning of life would have literally gotten me a I'm a language model and can't answer that for you I'm sorry kind of response
The usefulness of intelligence is in solving problems. Anything else is a sophisticated parrot. But a lot of the time we use AI for processing data instead of anything more novel. Turing test is so boring these days. It is just about lying after all.
These A.Is sometimes fail in simple logic tests. We are nowhere near AGI😂
Compare it to the average human G I - we're way past that level. I'm sure however, you never make a mistake.
I understand why the Ai didn't interpret your rain question as sarcasm. it wasn't a strong enough case for sarcasm. it sounded closer to how the ai interpreted it to me. So you need to come up with better scenarios 😮.
Feedback always welcome!
The usefulness of intelligence is in solving problems. Anything else is a sophisticated parrot. But a lot of the time we use AI for processing data instead of anything more novel. Turing test is so boring these days. It is just about lying after all.