Americans' Use of ChatGPT Is Ticking Up 25
Pew Research: It's been more than a year since ChatGPT's public debut set the tech world abuzz. And Americans' use of the chatbot is ticking up: 23% of U.S. adults say they have ever used it, according to a Pew Research Center survey conducted in February, up from 18% in July 2023. The February survey also asked Americans about several ways they might use ChatGPT, including for workplace tasks, for learning and for fun. While growing shares of Americans are using the chatbot for these purposes, the public is more wary than not of what the chatbot might tell them about the 2024 U.S. presidential election. About four-in-ten adults have not too much or no trust in the election information that comes from ChatGPT. By comparison, just 2% have a great deal or quite a bit of trust.
Most Americans still haven't used the chatbot, despite the uptick since our July 2023 survey on this topic. But some groups remain far more likely to have used it than others. Adults under 30 stand out: 43% of these young adults have used ChatGPT, up 10 percentage points since last summer. Use of the chatbot is also up slightly among those ages 30 to 49 and 50 to 64. Still, these groups remain less likely than their younger peers to have used the technology. Just 6% of Americans 65 and up have used ChatGPT. Highly educated adults are most likely to have used ChatGPT: 37% of those with a postgraduate or other advanced degree have done so, up 8 points since July 2023. This group is more likely to have used ChatGPT than those with a bachelor's degree only (29%), some college experience (23%) or a high school diploma or less (12%).
Most Americans still haven't used the chatbot, despite the uptick since our July 2023 survey on this topic. But some groups remain far more likely to have used it than others. Adults under 30 stand out: 43% of these young adults have used ChatGPT, up 10 percentage points since last summer. Use of the chatbot is also up slightly among those ages 30 to 49 and 50 to 64. Still, these groups remain less likely than their younger peers to have used the technology. Just 6% of Americans 65 and up have used ChatGPT. Highly educated adults are most likely to have used ChatGPT: 37% of those with a postgraduate or other advanced degree have done so, up 8 points since July 2023. This group is more likely to have used ChatGPT than those with a bachelor's degree only (29%), some college experience (23%) or a high school diploma or less (12%).
Re: (Score:2)
It's more that this is the group that is the most curious about it and that also may have questions where you can't easily google the answers.
Re: (Score:1)
Unfortunately ChatGPT gives incorrect answers, for things that should be relatively easy for it to figure out.
For example I was having trouble with converting some mesh data from Blender's format which has Y forward, X Right and Z up, to Unreal's format which has X forward, T right and Z up.Translation was easily fixed with swapping X and Y axis values, but rotations defined with euler angles were getting messed up and my model which was comprised of multiple sub meshes was ending up looking like a teleport
Re: (Score:2)
Naw, a coworker uses it, but we look over his shoulder and all laugh at the results. Like using a web search but wanting to see humorous and bizarre results. Highly educated adults are probably least likely to be affected by it is my guess. Current ChatBots really can't replace high skilled jobs. If it can effectively replace your job, then maybe you weren't as highly educated as you thought.
Trying vs Using (Score:5, Insightful)
Re: (Score:1)
Re: (Score:2)
My guess as well. I have noticed that even my 2nd semester programming students use it now only as a last resort (except for a few that will fail because they still cannot do elementary things themselves). They mentioned that even on only slightly more difficult questions, they often get unusable answers that look great but cause more effort to fix them than doing the work themselves.
Re: (Score:2)
I've found exactly the same in my field (tax law). Chat GPT gives a lot of outputs that look pretty good at first glance, but are dangerously wrong (especially if the person using Chat GPT isn't already an expert in the field). You can fix it yourself, but it can be more tedious to track down the falsehoods and fix the references than just drafting it yourself.
That said, it does have some uses. It's great for stuff like summarizing, translating, and simple data analysis (i.e. which terms are present in data
Re: (Score:2)
Interesting. The these two applications could not be much more different. That means the approach used is fundamentally broken for many applications and using it in an area as a non-expert is indeed dangerous.
I do not doubt it has uses. It can serve as a more capable search engine as well. But the defects will probably give it negative worth in the hands of many people.
Re: (Score:2)
My guess as well. I have noticed that even my 2nd semester programming students use it now only as a last resort (except for a few that will fail because they still cannot do elementary things themselves). They mentioned that even on only slightly more difficult questions, they often get unusable answers that look great but cause more effort to fix them than doing the work themselves.
Fair points, but the same could be said for textbooks. Neither textbooks nor ChatGPT (and let's throw stackexchange in for kicks) will help clueless students. All these things are resources. Expecting more (such as solving entire problems rather than pointing the way to limited solutions for specific parts of the problem) is a problem with expectations and not necessarily the resource.
Re: (Score:2)
It is a problem with the resource, because ChatGPT (and the like) confidentally pretend to be able to solve more complex problems (well, problems if tiny values for "more complex"). One of the mist important aspects of any engineer, technician and craftsperson is that thy know their limits. LLMs are a total failure in that regard. I call not knowing your limits "incompetence of the 2nd order", and it is a massively worse problem than plain, direct, known incompetence. Because with that people go for help. T
Re: Trying vs Using (Score:2)
Re: (Score:3)
Is the use of ChatGPT ticking up, or is the number of people who have given it a try ticking up?
Speaking strictly for myself I find I've been using ChatGPT more and more for searching various things, and also for quick and dirty scripts, minor debugging and the like.
For example, I got ChatGPT to write a Powershell script that downloads some files, parses some of their content then renames the files accordingly and pushes them to appropriate folders. I'm don't often use Powershell and I'm not interested in learning Powershell scripting so I couldn't write the script myself without spending quite a bit
Re: (Score:2)
Pretty much the same use-cases here. It is pretty impressive how far it has come already and we are just at the beginning.
Re: (Score:2)
Still, you'd probaby get equal or better results by using Google and selection the top non-advertisement result. That's all ChatBot is doing, except it mumbles up the words.
Re: (Score:2)
You've clearly not used either of those recently. Even with all the hallucinations, ChatGPT4 and Gemini Advanced responses are 10x quicker and more useful than digging through Googles trash search results that may or may not have anything to do with your query.
Stop talking out of your ass and actually try.
Re: (Score:2)
you'd probaby get equal or better results by using Google and selection the top non-advertisement result. That's all ChatBot is doing, except it mumbles up the words
It looks like you haven't ever used ChatGPT in any kind of depth, and talk in complete ignorance, but ok; maybe you're right. So let's test your assertion.
I described my use case above. Go ahead and use Google, take the top non-advertising result and show us a site that has a Powershell script that does what I described. Not separate fragments on multiple sites, mind - I don't want a link to some page with a script that shows how to download a file with Powershell, another link to some guy's video tutorial
My usage is going down not up. It's almost zero. (Score:2)
I remember a lawyer once showing me his law library and he said "This
Re:My usage is going down not up. It's almost zero (Score:5, Insightful)
I was just as interested as most folks when it came out. I used it and even found it helpful in writing technical documentation. I still use it for that occasionally. However, it fails at writing C code that will compile most of the time, refuses to try to write Assembler in most cases (and often fails badly when it does try) and thus isn't much help with $dayjob. I already canceled my OpenAI paid account. ChatGPT isn't worth it to me. I remember a lawyer once showing me his law library and he said "This is why we make the big bucks. I know almost every book and law in here." I was suitably impressed. Now, I expect folks will feed that law library to an LLM plus a lot more and the answers coming from the LLM would be somewhat on par with a real lawyer, minus the practical parts of his experience (ie.. which judges are friendly, where the court bathrooms are, etc..). If I had a job like that, I'd probably be somewhat nervous. However, looks like AI is oversold and underpowered compared to the fear we have around it.
most of us don't fear the AI itself. We fear the stupidity of people buying the AI hype. However strangely small that may look from a distance, that's a world of difference. Basically, we're scared of shitty management making shitty decisions about processes that will only get worse by sprinkling AI fairy dust over them because some sales dork was over-excited at a presentation.
Re: (Score:1)
Re: (Score:2)
The ChatGPT does NOT know how to program. At all. If the code snippets are not in its data base or not available on the web then it will not be able to give a suitable answer, though it may try. How could it know? I think there are a lot of gullible people out there who think it is intelligent or who do not know how it works. More techie mumbo jumbo, except it's fooling more educated people than past mumbo jumbo.
While I've only merely "tried" ChatGPT (Score:2)
Brave Search's AI Code LLM has been great for my use case. I keep extolling how great it is when I need to find something from stackoverflow and it ends up just creating it for me right there on the search page without having to parse a bunch of different links. This is probably irrelevant to the story but my use of a "AI chat bot" is definitely up order's of magnitude over the past few months.