This: don’t use AI to replace you, use it as a Texas sharp shooter to answer questions you have for which you’d otherwise used a www search engine.
I personally use ChatGPT as I used Google 20 years ago to find out about stuff a quick way. That’s my personal ELI4 (explain like I’m 4 yo). It’s also perfect to ingest an appliance manual and just do the tl;dr for you. Also, when looking to purchase stuff, it’s much better than Google and also does not propose sponsored contents.
I tried vibe coding an app and it worked but it was tedious to have so many recursion tests to perform as 4o will systematically add bugs alongside new functionalities. In this matter, I’d say it makes me better at troubleshooting, which is what my job consists of BTW.
Also prompting an AI is a skill: one has to seriously address it. The omg lol texting generation will not feel in power there.
Before we warn people about AI and resulting brain dysfunctions, I think the screen time of the youth as well as it’s connectivity should be addressed, because it already made too many victims.
Tl;dr: ChatGPT can be cool and fun, even useful, but it doesn’t replace a mind of one’s own. Use it, but not for more than you actually need it. Keep your brain on everytime lest that sneaky program might brainfsck you.🤓
BTW: who sponsored this study? Was the sample population adequate? What was the error rate?
Incorrect. Humans can not process data at AI speeds. AI is necessary for human development to continue past a point. And now may be a good time for everyone to look up the meaning of the word ARTIFICIAL. It does not mean fake.
I understand this statement given the intentional degradation of the Google search engine and the manipulation and censoring of output.
However, I would strongly disagree with the statement. A skilled user who understands how to ask questions, leverage personas and avoid AI hallucinations in the output can get results that dramatically exceed anything a search engine can provide.
Now that being said, I am not saying AI is a net positive. I am not at all surprised by the clinical trial referenced in this post and it's exactly why I limit my use of AI. I have colleagues that literally allow AI to write their business emails and that seems like a great way (to me at least) for one's brain to turn into mush eventually.
But regardless, there's an enormous delta between the output of AI and a search engine, as long as the user is reasonably proficient in engaging with the AI.
Well, it depends what you use a search engine for, or for that matter, AI.
AI is great at giving short answers to simple questions by querying the carefully curated walled garden that google has become. If that's your end goal, AI will get you there faster, to be sure.
If, instead, you're using a search engine to begin a journey of discovery or knowledge building and want options and differing viewpoints - something you could have gotten from Google 20 years ago, then, right now, AI isn't going to help you.
Or God forbid you just want to explore with no real goal in mind - some of my most interesting times on the web were exploring links that got pulled up and really weren't pertinent to my search but that caught my eye and sent me down a rabbit hole. That doesn't happen when AI (and google) is the gatekeeper to the results. Sometimes efficiency sucks all the joy out of life.
It's also a good idea to ask AI questions you already know the answer to. Watching it squirm and wriggle to justify lying to you is quite comical and scary at the same time. Think Killer Clowns from Outer Space.
This is no surprise. And, it is no coincidence the leading AI companies push the idea that users should allow AI to perform their writing tasks for them. Writing is not just a critical communication skill, it is also a powerful brain-building exercise.
It reminds me of a quote from the Star Trek Next Generation movie Insurrection:
We believe that when you create a machine to do the work of a man, you take something away from the man.
If you take away the need to improve writing skills then you effectively stunt a person’s continuing cognitive development. You take away their ability to think critically and grow as a person.
Surprise surprise if you have no common sense and are naturally dumber AI doesn't magically make you smarter. Apparently the opposite is true. I've been saying for a while it a good tool. That's it. It's not gonna write anything for you in some meaningful way, nor is it always correct. But using it to look up information on a given topic and summarize it is an EXCELELNT way to get a starting point on any given topic (just double check its all true and don't be afraid to push back if something just doesn't sound right, it's not impossible that it occasionally get something wrong, I've had it happen before and when I pushed back it DID correct itself and redo it with the correct information).
It's also an excellent calculator. That's my main use for it. Instead of having to type out dozens of formulas into an excel spreadsheet and fill it out, or have a dozen line of math formulas in a scientific/graphing calculator, you can just give the scenario, the starting numbers, and any other relevant information and let it work it all out for you. Again, its still good to double check because it has been known to get something wrong because of how you worded it on occasion, but as a calculator, it's top tier and simplifies advanced and long winded math that no one wants to actually sit down and do. That's what I do, mainly for future projections and financial feasibility of various projects. Saves me SO much time having to search public databases for valuation and doing projections for X amount of years, etc.
If you tell it to cite sources you can easily double check to make sure what its saying is correct and correct any errors it may have mad. You can also make sure its using up to date information as a reference and not something from 20 years ago, etc. etc.
Overall, it's a tool. It's literally just the world's fanciest calculator/search engine combo that you can talk to for therapy (which apparently people do). That's it.
Mech Eng/Programmer...PERSONALLY, I would rather do the research and come up with the equations or program myself than to use AI...Espousing AI as a "tool" IS NOT SOMETHING that should be undertaken lightly WITHOUT KNOWING THE CONSEQUENCES...The HUMAN MIND is a wonderful thing to use and we, as humans, were made to ask questions and find the answers ourselves. I studied AI when I was in college for Computer Science and it when programmed it was still "GARBAGE IN, GARBAGE OUT"!!!!!!
For the record, I'm more referring to financial projects. Think like 10 year revenue projections for a product, while accounting for multiple factors like downturns competition, etc. etc. Things that normally have a margin of error and you'd redo a normal, high and low projection anyway. I have no idea about engineering, but as far as my purposes, as long as you give it the right information it's pretty accurate within the acceptable margin of error.
Took some trial and error (which is where the tool part comes in, since you can't just blindly trust it), since at first it would give me crazy financials like 700% ROI on something simple like ROI projections for a restaurant, but after refining my "verbiage formula" I'll call it and figuring out how to upload statistical info and make it cite what its using that I don't give it myself, It's an excellent tool.
Like I said, its a tool. No different from google searching or a Texas Instruments Calculator. Just more advanced. If you know how to use it and refine it, it works excellent for its purpose. If you just blindly trust it then you get REALLY stupid responses.
For example, there was a semi viral video a while back of some black woman asking ChatGPT what would happen if all the black people left to africa as far as the economy goes and it was ridiculous crap like 7 trillion dollars in financial activity would shift from the US to Africa, and crap like that. Didn't account for any extraneous factors or broader geopolitical and economic factors either.
I stand by what I said. It's a good tool, but like any tool, if you're dumb and just blindly trust it, then you'll obviously have problems. For anything important you'd always double check it. I'll often ask it for general information (with citations) and then build up the information gathering myself from there while confirming it in a different tab, and then once I've found all the information I need, just frontload the question in a new chat (since chat degradation is a thing), and refine the data with back and forths until I'm satisfied that its within an acceptable margin of error that logically makes sense, and then go and confirm it myself
It is seductive. The instant data collation is so useful, so is the ability to quickly write boring officialise crap, but I am noticing that I'm using it instead of Google searching for stuff. I can feel the temptation to use AI instead of working stuff out for myself. Fortunately the very high incidence of AI hallucination keeps me very cynical about AI. I find whatever efficiency savings AI makes are eroded by how much verification I have to do.
And people don't understand why I get on the people who are constantly "Asking Grok" like its supposed to be insightful or something. It just made the poster asking it lazy.
You'll find the same in every person who's ever just copied and pasted from books for their essay. It's not like gpt's copy and paste is new to students who get the "write five pages on the various qualities of dirt" level assignments given by checked out teachers.
Have them use gpt for writing about something they care about. That engagement will have them double checking GPTs answers and will change the scan map dramatically.
This: don’t use AI to replace you, use it as a Texas sharp shooter to answer questions you have for which you’d otherwise used a www search engine.
I personally use ChatGPT as I used Google 20 years ago to find out about stuff a quick way. That’s my personal ELI4 (explain like I’m 4 yo). It’s also perfect to ingest an appliance manual and just do the tl;dr for you. Also, when looking to purchase stuff, it’s much better than Google and also does not propose sponsored contents.
I tried vibe coding an app and it worked but it was tedious to have so many recursion tests to perform as 4o will systematically add bugs alongside new functionalities. In this matter, I’d say it makes me better at troubleshooting, which is what my job consists of BTW.
Also prompting an AI is a skill: one has to seriously address it. The omg lol texting generation will not feel in power there.
Before we warn people about AI and resulting brain dysfunctions, I think the screen time of the youth as well as it’s connectivity should be addressed, because it already made too many victims.
Tl;dr: ChatGPT can be cool and fun, even useful, but it doesn’t replace a mind of one’s own. Use it, but not for more than you actually need it. Keep your brain on everytime lest that sneaky program might brainfsck you.🤓
BTW: who sponsored this study? Was the sample population adequate? What was the error rate?
The sad thing, is that if we had the Google of 20 years ago, we wouldn't have a fraction of the need for AI to search for us.
Google has created the problem and then spent a fortune developing the "solution".
Back then I was a huge fan of AltaVista.
I miss Ask Jeeves
Yandex is like the old google used to be, but the capchas are too much.
Incorrect. Humans can not process data at AI speeds. AI is necessary for human development to continue past a point. And now may be a good time for everyone to look up the meaning of the word ARTIFICIAL. It does not mean fake.
Now might be a good time to look up the word "intelligence" because AI ain't it.
This. (And if I knew how to increase the font size, put in italic caps and color it bright red in bold, I would do it)
I understand this statement given the intentional degradation of the Google search engine and the manipulation and censoring of output.
However, I would strongly disagree with the statement. A skilled user who understands how to ask questions, leverage personas and avoid AI hallucinations in the output can get results that dramatically exceed anything a search engine can provide.
Now that being said, I am not saying AI is a net positive. I am not at all surprised by the clinical trial referenced in this post and it's exactly why I limit my use of AI. I have colleagues that literally allow AI to write their business emails and that seems like a great way (to me at least) for one's brain to turn into mush eventually.
But regardless, there's an enormous delta between the output of AI and a search engine, as long as the user is reasonably proficient in engaging with the AI.
Well, it depends what you use a search engine for, or for that matter, AI.
AI is great at giving short answers to simple questions by querying the carefully curated walled garden that google has become. If that's your end goal, AI will get you there faster, to be sure.
If, instead, you're using a search engine to begin a journey of discovery or knowledge building and want options and differing viewpoints - something you could have gotten from Google 20 years ago, then, right now, AI isn't going to help you.
Or God forbid you just want to explore with no real goal in mind - some of my most interesting times on the web were exploring links that got pulled up and really weren't pertinent to my search but that caught my eye and sent me down a rabbit hole. That doesn't happen when AI (and google) is the gatekeeper to the results. Sometimes efficiency sucks all the joy out of life.
If you read the full report, you will find exactly this. The above article conveniently omits much of what the study found.
It's also a good idea to ask AI questions you already know the answer to. Watching it squirm and wriggle to justify lying to you is quite comical and scary at the same time. Think Killer Clowns from Outer Space.
Never met a Texas Sharp Shooter. Kek
Here: https://academic.oup.com/humrep/article/32/7/1363/3852142
WOAH. Time to pump the breaks! Those are some seriously worrisome statistics!
look at the authors
it's controlled
https://nitter.poast.org/itsalexvacca/status/1935343874421178762
From the end of the thread:
This is no surprise. And, it is no coincidence the leading AI companies push the idea that users should allow AI to perform their writing tasks for them. Writing is not just a critical communication skill, it is also a powerful brain-building exercise.
It reminds me of a quote from the Star Trek Next Generation movie Insurrection:
If you take away the need to improve writing skills then you effectively stunt a person’s continuing cognitive development. You take away their ability to think critically and grow as a person.
Surprise surprise if you have no common sense and are naturally dumber AI doesn't magically make you smarter. Apparently the opposite is true. I've been saying for a while it a good tool. That's it. It's not gonna write anything for you in some meaningful way, nor is it always correct. But using it to look up information on a given topic and summarize it is an EXCELELNT way to get a starting point on any given topic (just double check its all true and don't be afraid to push back if something just doesn't sound right, it's not impossible that it occasionally get something wrong, I've had it happen before and when I pushed back it DID correct itself and redo it with the correct information).
It's also an excellent calculator. That's my main use for it. Instead of having to type out dozens of formulas into an excel spreadsheet and fill it out, or have a dozen line of math formulas in a scientific/graphing calculator, you can just give the scenario, the starting numbers, and any other relevant information and let it work it all out for you. Again, its still good to double check because it has been known to get something wrong because of how you worded it on occasion, but as a calculator, it's top tier and simplifies advanced and long winded math that no one wants to actually sit down and do. That's what I do, mainly for future projections and financial feasibility of various projects. Saves me SO much time having to search public databases for valuation and doing projections for X amount of years, etc.
If you tell it to cite sources you can easily double check to make sure what its saying is correct and correct any errors it may have mad. You can also make sure its using up to date information as a reference and not something from 20 years ago, etc. etc.
Overall, it's a tool. It's literally just the world's fanciest calculator/search engine combo that you can talk to for therapy (which apparently people do). That's it.
Mech Eng/Programmer...PERSONALLY, I would rather do the research and come up with the equations or program myself than to use AI...Espousing AI as a "tool" IS NOT SOMETHING that should be undertaken lightly WITHOUT KNOWING THE CONSEQUENCES...The HUMAN MIND is a wonderful thing to use and we, as humans, were made to ask questions and find the answers ourselves. I studied AI when I was in college for Computer Science and it when programmed it was still "GARBAGE IN, GARBAGE OUT"!!!!!!
For the record, I'm more referring to financial projects. Think like 10 year revenue projections for a product, while accounting for multiple factors like downturns competition, etc. etc. Things that normally have a margin of error and you'd redo a normal, high and low projection anyway. I have no idea about engineering, but as far as my purposes, as long as you give it the right information it's pretty accurate within the acceptable margin of error.
Took some trial and error (which is where the tool part comes in, since you can't just blindly trust it), since at first it would give me crazy financials like 700% ROI on something simple like ROI projections for a restaurant, but after refining my "verbiage formula" I'll call it and figuring out how to upload statistical info and make it cite what its using that I don't give it myself, It's an excellent tool.
Like I said, its a tool. No different from google searching or a Texas Instruments Calculator. Just more advanced. If you know how to use it and refine it, it works excellent for its purpose. If you just blindly trust it then you get REALLY stupid responses.
For example, there was a semi viral video a while back of some black woman asking ChatGPT what would happen if all the black people left to africa as far as the economy goes and it was ridiculous crap like 7 trillion dollars in financial activity would shift from the US to Africa, and crap like that. Didn't account for any extraneous factors or broader geopolitical and economic factors either.
I stand by what I said. It's a good tool, but like any tool, if you're dumb and just blindly trust it, then you'll obviously have problems. For anything important you'd always double check it. I'll often ask it for general information (with citations) and then build up the information gathering myself from there while confirming it in a different tab, and then once I've found all the information I need, just frontload the question in a new chat (since chat degradation is a thing), and refine the data with back and forths until I'm satisfied that its within an acceptable margin of error that logically makes sense, and then go and confirm it myself
AI can be and is often cometely wri g. You better back up anything it tells you with a second reliable source.
It is seductive. The instant data collation is so useful, so is the ability to quickly write boring officialise crap, but I am noticing that I'm using it instead of Google searching for stuff. I can feel the temptation to use AI instead of working stuff out for myself. Fortunately the very high incidence of AI hallucination keeps me very cynical about AI. I find whatever efficiency savings AI makes are eroded by how much verification I have to do.
And people don't understand why I get on the people who are constantly "Asking Grok" like its supposed to be insightful or something. It just made the poster asking it lazy.
Digital Cliff Notes- Call it whatbit really is- MACHINE LEARNING, to learn they must be programmed....by who???
You'll find the same in every person who's ever just copied and pasted from books for their essay. It's not like gpt's copy and paste is new to students who get the "write five pages on the various qualities of dirt" level assignments given by checked out teachers.
Have them use gpt for writing about something they care about. That engagement will have them double checking GPTs answers and will change the scan map dramatically.