Rajeev Tyagi

I was great with routes in my youth.
I started using Google Maps, no longer required to use that skill, became dull at it.

Similarly, with #GenerativeAI and engaging with LLMs, I foresee less human interaction, especially fewer discussions on urgent matters—another skill I excelled at.
#EduSkyAI

To the founders/owners of schools:
If you start a school, then primarily the child comes to you.

You hire more people to help you do your work, and you are only responsible for ensuring that they deliver what you hired them for.

#EduSky

issue using #GenerativeAI for exam preparation-

Mirror eqn written by the student-
1/v + 1/u = 1/f
GenAI ( #DeepSeek)-
the eqn is 1/f = 1/u + 1/v
->student has interchanged v and f
->answer is incorrect!
=>GenAI failed to see the two expressions are same, ALSO gave a WRONG answer
#EduSky #EduSkyAI

#ChatGPT's memory feature offers insights into your user profile under Settings > Personalization > Memory > Manage.

While I find the profile it generated quite accurate, it could be improved if the data were editable. Currently, the only option available is to delete these remarks entirely.

This means, I am looking for my own name

Edit,

my dear @bsky.app

because we humans are not perfect.

We realise we wanted to say something different 'after' we post!

Looking for someone in this list who has the same name as mine
😀

Genes.

C'mon Champ!!!

How would rajeevtyagi.twitter.social feel like in place of a simple rajeevtyagi as my handle?

Then why does @bsky.app think otherwise!

C-to-C...

Students using
chatGPT to write assignments

Teachers using
chatGPT to write comments while checking.😆

#ArtificialIntelligence #EdTech #EduSkyAI

Another request -
Kindly add "following since", so that those whom I expect to follow back, but they do not, can be unfollowed!
@bsky.app

Whao! DONE!!!
👍

This one is seriously good stuff!!!
#DeepSeek
because it shows you what it thought through (before it answered) you can hit at those points where you differ in the logic itself

for one question, it thought for 18 seconds, and revealed all that was going on in its thoughts for those 18 secs!

Tried DeepSeek today (switched on the "deep think" slider, it gives 50 free messages at that level, free otherwise).

Liked how it just opens up its mind (revealing the thinking process it is going through)

just this that the problem should be worth giving a deep thought

#NotebookLM does much better work 'for free' when you upload one chapter at a time, the previous years' questions, and ask it to prepare the child to take an exam based on them.

@bsky.app
If the mouse hovers over a name, let a brief bio appear with an option to follow the person.
Right now, one must click, open the page, and then click on follow.

Over 1,15,000 users leave X after Trump's win After the 2024 United States Presidential election that confirmed Donald Trump's victory, social media platform X (formerly Twitter) experienced its largest user exodus since Elon

I am not part of the Twitter/BlueSky debate, I just want to connect to fellow teachers, from whom I have much to learn!

#EduSky

m.rediff.com/news/comment...

#edusky
How do I convince you I am an educator?

Just joking, actually found this amusing!🙂

Phew!!!! Let me breathe.
If you are interested, I will continue to share my journey with you. 😊

#GenerativeAI #ChatGPT4o

Requesting it to develop a lesson plan to take a fresher to the level where the student can solve the kind of problems asked in previous years. Expecting it to look for appropriate videos, simulations and ask thoughtful questions at each stage to ascertain learning has happened.

Initially I asked it to look for previous years' questions from the internet, but then provided the same to be sure that the 'agent' knows the pattern well.

After pleading the 'agent' in many ways (prompt engineering!!!) to ask the student few fundamental questions to check the pre-knowledge required to study that chapter, I finally provided a list of questions I consider appropriate.

Stage 1: (Frustration) -
How do you talk to a super-hyped worker ("I am an IITian" type), who is a big yes-boss type, aligning opinion more quickly than the boss changing his/her own opinions, and who just keeps giving suggestions without implementing any!

Sharing my experience of developing a customised GPT, that can teach you one chapter of Physics (at a time).

Those who know this topic can see that even the conclusion is wrong.

And, it is wrapped in layers and layers of what appears to be the "pearls of wisdom"

#generativeAI #AIEd

While the sources don't detail all these scenarios, they provide the foundation for understanding image formation in concave mirrors. Specifically, the mirror formula, ray diagrams, and the sign conventions are crucial in understanding how image characteristics change with varying object positions

But after a lot of junk explanation, it concluded with this -
NotebookLM (pasting the response as received):
Therefore, there isn't a single correct answer for Question 1 as it is currently phrased. A more accurate question would specify the object's initial position relative to the concave mirror.

Me: Your answer to Question 1 is wrong. Can you give me the source from where you got this answer. Also explain why you think your answer is correct.
NotebookLM gave a long descriptive answer, seemingly explain all possible scenarios. I will attach that in comments if anyone is interested.

A. Move closer to the mirror and become smaller.
B. Move away from the mirror and become larger.
C. Move closer to the mirror and become larger.
D. Move away from the mirror and become smaller.

I asked it to provide the answer key. It said: "C. Move closer to the mirror and become larger".