Shulou(Shulou.com)11/24 Report--

Recently, Harvard announced a major decision: AI mentors will enter the course.

He is also in charge of the flagship project of the computer department, introduction to computer Science, also known as CS50.

With robot mentors, Harvard's CS50 program will have a teacher-to-student ratio of 1:1.

The news was mentioned in an email by Professor David Malan, the mentor of the CS50 program, and the semester will officially start this fall.

This AI is called CS50 bot, and its core is the large language Model (LLM).

However, instead of using ChatGPT, Bard and other popular products on the market, it is developed by Harvard itself.

CS50 bot can answer students' questions, and can also provide services such as code debug, project analysis and so on.

With the help of CS50 bot, in addition to the teacher-student ratio, Harvard hopes to achieve the following visions:

Give students 24-hour help for 7 hours.

Provide personalized guidance based on learning habits

Targeted teaching to narrow the gap between students

Adapt to the learning level and progress of each student

Generate customized exercises and exam questions

Maintain students' critical thinking despite the AI tools, Professor Malan stressed the importance of critical thinking.

We will make it clear to students that they should always think critically when receiving information, whether it comes from humans or software.

And AI leads to the decline of critical ability, which is exactly what many people are worried about.

However, in addition to reminding students, Harvard's CS50 bot is also designed to avoid this problem.

The reason why Harvard does not directly use LLM products on the market is precisely because these LLM are "too powerful" in the eyes of scholars.

AI chat tools such as ChatGPT can directly give answers to questions, which will breed students' inert thinking over time.

In response to this, Harvard's CS50 bot specially carried out a "negative optimization"--

It does not answer students' questions directly, but guides students to think and teach them how to find answers on their own.

Or when a student submits a piece of code, bot helps with the debug.

In short, bot will only come in handy when you really need help, and it won't work if you want to be lazy.

Belated AI Policy although Harvard developed its own AI teaching machine relatively early, there has been no policy on AI teaching before.

The exact reason is not known.

This AI teaching is also a small-scale experiment carried out in the CS50 project.

It can be seen that Harvard is more cautious on the issue of introducing AI.

In fact, the debate about AI teaching never stops, and the practices of other schools are not the same.

Some are banned altogether. For example, Steve West, Vice Chancellor of the University of West England, warned students not to try to use ChatGPT to finish their homework.

You are smarter than AI. You don't need to use these tools.

Steve also said that although it is not possible to bring the established technology (AI) back to before it was invented, it must be limited.

There are also some colleges and universities that are open to AI.

For example, the University of Washington has fully liberalized the use of AI, even as long as it is explained, it doesn't hurt to use it in exams.

A law school professor at the University of California, Los Angeles, also said in the article that he fully allows students to use AI for writing tasks.

Another common practice is to allow it to be used conditionally.

At Cambridge University, for example, students are not allowed to use AI for exams or writing assignments, but are free to use it in other cases.

The University of Sydney leaves it up to teachers to decide whether to allow students to use AI tools.

These policies have their own considerations, and it is impossible to simply judge whether they are good or bad.

Back to Harvard, from the netizens' comments, this approach is more recognized.

This netizen believes that the development of AI is unavoidable for human beings, so we can only face it squarely. Even if educators issue a policy prohibiting students from using it, some students will steal it, and they are unwilling to admit that it is better to guide students to use AI correctly and let them get help from AI aboveboard.

Reference link:

[1] https://www.independent.co.uk/tech/harvard-chatbot-teacher-computer-science-b2363114.html

[2] https://www.thecrimson.com/article/2023/6/21/cs50-artificial-intelligence/

[3] https://www.reddit.com/r/singularity/comments/14k5141/harvards_new_computer_science_teacher_is_a_chatbot/

[4] https://news.ycombinator.com/item?id=36452457

This article is from the official account of Wechat: qubit (ID:QbitAI). Author: Creasy.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.