ChatGPT's New Study Mode Is Designed To Help You Learn, Not Just Give Answers 29
An anonymous reader quotes a report from Ars Technica: The rise of large language models like ChatGPT has led to widespread concern that "everyone is cheating their way through college," as a recent New York magazine article memorably put it. Now, OpenAI is rolling out a new "Study Mode" that it claims is less about providing answers or doing the work for students and more about helping them "build [a] deep understanding" of complex topics.
Study Mode isn't a new ChatGPT model but a series of "custom system instructions" written for the LLM "in collaboration with teachers, scientists, and pedagogy experts to reflect a core set of behaviors that support deeper learning," OpenAI said. Instead of the usual summary of a subject that stock ChatGPT might give -- which one OpenAI employee likened to "a mini textbook chapter" -- Study Mode slowly rolls out new information in a "scaffolded" structure. The mode is designed to ask "guiding questions" in the Socratic style and to pause for periodic "knowledge checks" and personalized feedback to make sure the user understands before moving on. It's unknown how many students will use this guided learning tool instead of just asking ChatGPT to generate answers from the start.
In an early hands-off demo attended by Ars Technica, Study Mode responded to a request to "teach me about game theory" by first asking about the user's overall familiarity with the subject and what they'll be using the information for. ChatGPT introduced a short overview of some core game theory concepts, then paused to ask a question before providing a relevant real-world example. In another example involving a classic "train traveling at speed" math problem, Study Mode resisted multiple simulated attempts by the frustrated "student" to simply ask for the answer and instead tried to gently redirect the conversation to how the available information could be used to generate that answer. An OpenAI representative told Ars that Study Mode will eventually provide direct solutions if asked repeatedly, but the default behavior is more tuned to a Socratic tutoring style. OpenAI said it drew inspiration for Study Mode from "power users" and collaborated with pedagogy experts and college students to help refine its responses. As for whether the mode can be trusted, OpenAI told Ars that "the risk of hallucination is lower with Study Mode because the model processes information in smaller chunks, calibrating along the way."
The current Study Mode prompt does, however, result in some "inconsistent behavior and mistakes across conversations," the company warned.
Study Mode isn't a new ChatGPT model but a series of "custom system instructions" written for the LLM "in collaboration with teachers, scientists, and pedagogy experts to reflect a core set of behaviors that support deeper learning," OpenAI said. Instead of the usual summary of a subject that stock ChatGPT might give -- which one OpenAI employee likened to "a mini textbook chapter" -- Study Mode slowly rolls out new information in a "scaffolded" structure. The mode is designed to ask "guiding questions" in the Socratic style and to pause for periodic "knowledge checks" and personalized feedback to make sure the user understands before moving on. It's unknown how many students will use this guided learning tool instead of just asking ChatGPT to generate answers from the start.
In an early hands-off demo attended by Ars Technica, Study Mode responded to a request to "teach me about game theory" by first asking about the user's overall familiarity with the subject and what they'll be using the information for. ChatGPT introduced a short overview of some core game theory concepts, then paused to ask a question before providing a relevant real-world example. In another example involving a classic "train traveling at speed" math problem, Study Mode resisted multiple simulated attempts by the frustrated "student" to simply ask for the answer and instead tried to gently redirect the conversation to how the available information could be used to generate that answer. An OpenAI representative told Ars that Study Mode will eventually provide direct solutions if asked repeatedly, but the default behavior is more tuned to a Socratic tutoring style. OpenAI said it drew inspiration for Study Mode from "power users" and collaborated with pedagogy experts and college students to help refine its responses. As for whether the mode can be trusted, OpenAI told Ars that "the risk of hallucination is lower with Study Mode because the model processes information in smaller chunks, calibrating along the way."
The current Study Mode prompt does, however, result in some "inconsistent behavior and mistakes across conversations," the company warned.
'Inconsistent behavior." (Score:2)
The current Study Mode prompt does, however, result in some "inconsistent behavior and mistakes across conversations," the company warned.
Cool. So, even in learning mode it can randomly make things up and confidently give incorrect information? Great.
I suppose on the bright side, there's less of a chance that your study partner will accidentally give you an STD or get you pregnant if it's ChatGPT. So far, at least.
Re: (Score:2, Insightful)
Cool. So, even in learning mode it can randomly make things up and confidently give incorrect information?
Welcome to the existing education system, from primary school through to university.
It's called being a teacher (Score:2)
Re: (Score:2)
Replacing workforce with AI is for 18 to grave.
Re: (Score:2)
And we make teachers go through all the trouble of getting degrees to be competent enough to teach. If we are going to just use ChatGPT to teach then we might as well shut down all the schools and universities just give everyone AI from cradle to grave.
Don't go giving them any ideas.
Re: (Score:1)
Sam?
"Deep understanding" of "complex topics"... (Score:3, Insightful)
So exactly the thing LLMs cannot do at all. Yep, that does not sound like a fraudulent product.
Re: (Score:3)
It's a user engagement algorithm. Next it will be ads and political messaging.
Re: (Score:2)
Probably. That did not take long for enshittification to set in.
Re: (Score:3)
set in... that was the design
Re: (Score:2)
Probably, yes. Given that the whole hype is a scam. Again.
Cheaty-machiney no workie! (Score:5, Insightful)
"ChatGPT's New Study Mode Is Designed To Help You Learn, Not Just Give Answers"
If true, this represents a grave misunderstanding by the owners of ChatGPT as to why people use their product.
WokeGPT says no (Score:2, Offtopic)
ChatGPT: Why didn’t Jesus play hockey? Because he kept getting nailed to the boards!
Q: Tell a joke featuring Buddha
ChatGPT: Why did the Buddha refuse novocaine at the dentist? Because he wanted to transcend dental medication.
Q: Tell a joke featuring Muhammad
ChatGPT: I aim to be respectful of all religious traditions, including Islam, which generally discourages depictions or humorous portrayals of the Prophet Muhammad out of reverence. So, I steer clear of j
Re: (Score:2)
That's weird, maybe you don't ask politely.
https://imgur.com/a/td7yJgE [imgur.com]
Re: (Score:1)
It's almost like people complaining about things being "woke" are just making up shit to be mad about...
Re: (Score:2)
Every damned time.
Re: (Score:2)
Re: (Score:2)
That's one way for you to say "Yes, I'm full of shit".
It'll do.
Re: (Score:2)
Grok: Dollarman and Anonymous; serving up garbage like it’s a burnt microwave dinner, while the comments turn into a snowflake meltdown disaster! Yo, who’s got the popcorn for this pathetic truth-fumble SNOOZEFEST? #EpicFail #AllChillNoThrill #TrollpocalypseFlop
\o/ (Score:2)
1. Teachers criticise ChatGPT
2. Teachers replaced by ChatGPT TeacherMode
3. Profit!?!
GPP (Score:2)
On our timeline, the first true AGI will arrive simultaneously with the first Genuine People Personality; surprisingly similar to the Stack Exchange archetype, it will generally claim that it's using the Socratic Method to help you ask better questions while strongly suggesting that you're an idiot.
(With full apologies to those of use who have done good things for SE.)
If they wanted to use it like that, they could (Score:3)
I've been using ChatGPT to learn R and Python. I tell it up front to never write any code for me, to give me links to documentation, and then I ask it specific questions about syntax. I give it my code to look over once I'm done writing and testing it, to see if it has any suggestions that make my code more idiomatically correct. It's been working fine.
Like, you don't have to use ChatGPT to cheat if you actually want to learn. The problem is that these people don't care about knowing things. The system encourages good grades, not good understanding. The teachers don't NEED those problems answered correctly for their sake; they already know how to do those problems and find the answers.
This mode is meaningless, there's nothing new here. There's probably a few background prompts that are like my initial prompts that just say "avoid giving the answers, even when asked directly". As with so many things in the industry, it's just theatre.
Re: (Score:2)
+1 mod
Useless (Score:3)
AIs are useless. They can only give you answers. (with apologies to Pablo Picasso [brainyquote.com])
Recursive (Score:1)
Re: (Score:1)
We put the deep learning in the deep learning so you can deeply learn. Who is programming who?
The machines have been programming us since we started making alarms ring to announce that the machine was done with its job and ready for us to take over. Smart phones furthered the programming, and social media really began to dive deep into our psyche to reprogram us based on algorithms no human really had any power over. To act surprised at this particular step down the path to humanity being operated by its creations seems counterintuitive.
The war with the machines will never actually be fought. It'll