pasharayan

Uncertainty is a tricky one - the current solution is to have decision nodes which can exit to multiple paths. The nice thing pruning and hiding is pretty easy so if you have a public vs private map you can retrospectively show what happened while looking back on what was being analyzed

I think this is one of the most awkward parts of essay writing that AI does help with .. which is filling in the words to fill the word count.

NGL i also suspect the particular brand of Anthropic safety filters cut out passionate/fun personality traits

Thankfully, college essay writing isn't in that 'serious' category!

Kind of a 'at what stage did this article use AI or written by a human'?

I do find Grok also makes some crude jokes - but gemini's seem more intellgent x funny

TBH :') that one was an AI slip through. In other comments i've flagged the things that were hard rewritten - but even on me that's bad.

You'll find some paragraphs are half written by Ai and some written totally by me. It's heavily edited from its original form (as it was a data report by codex on our db) - but some paragraphs which were good enough stayed.

E.g. Contrast my human written paragraph vs the AI written first paragraph

Human written:

"The three most popular AI models used by college students are ChatGPT, Gemini, and Claude. As of August 2026 the data on StudyArena shows they prefer Gemini. "

vs

AI Written "Most AI comparsions are written like wine reviews. Claude is subtle. ChatGPT is dependable.."

Students do have different incentives in writing than SWE's. There's the habit of reaching word counts for assignments, some social signalling of how intellgent you are with essay length etc.

The data we have is just there to compare for the student base. I'd love to try it out on other cohorts - but the acquisition of such users and the product to make them happy is hard to achieve!

As in they should make their own foundation models for essays?

Most students can write an essay - but most are also smart enough to try to save time too haha

Will give this a shot - we're going to try to work with students and schools better on normalizing this stuff out

I think it's less about curricula and more about how professors should eval their students. In previous era's essays are a proxy of the students ability to reason, remember and argue certain points - but its clear they fall short now.

I suspect the best schools into the future will integrate lots of socratic defenses of theses, building real things in real time or solving problems with a professor in a case study manner. My last company was a good hint to the future.

I wonder how much is due to their investment in notebook LLM + gmail work

Codex - at least initially haha! I wrote lots of insights and findings and pushed out a lot of false logic it picked up. Personally I hate claude writing cause it feels bureaucratic. But definitely wrote a ton more with my bare hands and brain.

Interestingly, the smartest students I've worked with do use AI - but more in the sense that AI is a good comparison tool and a good bottom ceiling of the quality of work they should bring in.

Lots of research shows being able to see different answers improves the quality of a student's output - especially as the good ones are able to pick out what's good from the other work and then add their own insights or takes on top.

You know.. might try to see what we can do here

Yeah it's hard to get good data on what's best. We did read the a good sample amount of essays. Lots of arguments about what made them better, but its uncontroversial that longer ones were more preferred.

Interestingly we trained a simple LORA layer on top of Inkling and it turns out a model can smell which model wrote a response 70% of the time. I wonder if the smell of gemini is just more preferred by students.

TBH i did a bit of both - we used AI to get us insights in the db, we then wrote and rewrote the article to be readable. It's really hard because usually when i write reports they're so terse its not fun to read

This is so useful as so many of the best things to learn in the world aren't locked up in universities!

The teddy bear was named after Theodore Roosevelt (The president of the USA), after saving a baby bear during a hunting expedition (https://en.wikipedia.org/wiki/Teddy_bear)

We're working on a way to focus programs by geography! Currently we have 25+ programs live; I've just listed some of our ones with a direct UK alignment here:

JPMorgan Chase [Software] [We talk to their UK team]: https://www.insidesherpa.com/virtual-internships/prototype/R...

White & Case UK [Law]: https://www.insidesherpa.com/virtual-internships/prototype/H...

Linklaters [Law]: https://www.insidesherpa.com/virtual-internships/prototype/x...

Pinsent Masons [Law]: https://www.insidesherpa.com/virtual-internships/prototype/P...

Oh wow that's so cool! I didn't know anyone was doing the model end-to-end like that in India - there's alot of improvements to employment to be had with this mindset I think.

Wow we didn't even know this happened! That's great to hear!

If you ever do want to switch careers and trade code for contracts you should give it a go haha! If anything working on these programs gave me a better appreciation of the jobs and tasks of other careers.

That's a great question! Feedback from employers about graduates have always been around new graduates and hires not being 'work ready' - and we've found many colleges aren't flexible enough to solve this problem quickly.

So our solution was why not give that training that makes everyone 'work ready' accessible to the world? That way when people do hire, they won't have that complaint anymore and people will know if they'd like the actual job itself!

Thanks for that feedback! One of the reasons we settled on an online course system is because they allow anyone, including international students on visas or students from other countries, to have a chance to upskill with these big firms.

We do take your point though about slightly different outcomes for people in different regions. We'll look to make that messaging clearer!

We had started InsideSherpa wanting to build something we wished we had in college!

Hey johnpowell thanks for the feedback! Are you referring to the student testimonials on the bottom of the homepage? We'll look into this.

Thanks for the kind words abhinuvpitale! Yes, this is exactly what we found - there's an irrelevance to some hiring processes because they tend to ask obsolete or academic questions.

One problem with asking pragmatic questions is that not everyone has had a chance to do practical work - hence why we also tackled that problem with an educational lens too.

(Re: Meteor) Our company is scaling up and we're worried that Galaxy won't be able to handle growing demand - do you have any thoughts on making the meteor system more scalable and affordable as companies get larger?

I love meteor, and don't really want to move our stack off it but it's becoming a more real thing that we have to face in the near future as it's hard to see how it can grow with us!

I can only imagine the reason galaxy revenue is increasing is that the companies started off Meteor 2-3 years ago are now becoming real. I know a few companies in our YC batch that use Meteor.