from Mike Johnson, about DOGE's use of algorithms: ""Elon has cracked the code. He is now inside the agencies. He’s created these algorithms that are constantly crawling through the data. And as he told me in his office, the data doesn’t lie. We’re going to be able to get the information."
My speculation, but it's very likely that one of those pieces of analyzed information was the 5 bullet points from each federal employee justifying their job.
I really don't think this is entirely spurious considering Elon has been using LLMs to fire government employees. It seems like a very plausible explanation for how they came up with this policy.
Funny that this is the exact calculation that ChatGPT suggests. Seems like another genius Musk idea. Madagascar definitely deserves 47% tariffs on US imports.
Must be hard to have to deal with that, I'm so sorry you have to use a different word when referring to someone.
It's obviously much harder for you to use a different pronoun than for them to have their entire gender identity invalidated. Would you like for people to misgender you constantly and purposefully?
Your life must be so hard. I feel bad for you to have to use a different word to refer to someone. Maybe have some basic decency and respect for others?
Genuinely, how hard for you is it to call a person what they want to be called. What difference does gender make in the workplace even?
I understand it's hard for you to use different pronouns, but it's probably much harder for the person who feels like they don't identify with their gender assigned at birth.
> Requiring me to call someone who was born with a penis, and who has sired children a "she" is telling me to say that 2 + 2 = 5.
It's more like someone telling you to call them John and you call them Tyler instead. It's literally just a word.
Love the idea! We're doing something similar to parse rubrics and student submissions at https://automark.io - great to see an open source library exploring the space more! Like you said, I think iteratively adding explicit layers of LLM understanding to the raw extraction will allow a lot more control over what information gets extracted. Also interested to see an integration with GPT-4V as an additional aid. I'd love to chat sometime if you have time - my email is in my bio.
As someone working on an AI tool for education, I think it's going to be one of the hardest industries to get right, but also the most important.
- It's going to be tricky for AI tutors to make an impact. Teachers are better at inspiring students to do better than computers ever will be, but for those already motivated, AI will provide a great study tool.
- Students have access to LLMs as well now, so any writing assessment will have to be timed and closely monitored, or in-class, otherwise students can easily cheat.
- Teacher-facing tools will come before student-facing tools because of the safety aspect.
We're building an AI grading assistant for teachers (https://automark.io) that helps them provide more writing feedback to students. Papers are extremely time consuming to grade and give feedback on, which is detrimental to both the teacher and the student. Giving students unlimited practice essays before the scored essay is one area where AI will really move the needle.
Grading is also one of the main shortcomings of online courses, so I'm guessing we'll see more in that area in the future where high quality courses can scale to 10k-100k+ students.
Same region, different provider. Also, their JS API has to do a round trip every time you set auth credentials on the backend, so it's at least 2 round trips for a single query.
- Now, the connection between our API servers and the database was slow (a few hundred ms per query), so we moved to self-hosting Postgres which was pretty painful. We tried to use https://github.com/supabase/postgres, but the documentation was very lacking and we had to make a bunch of modifications to get it to work. After we got it working, it was pretty smooth though - pretty easy to implement backups, etc.
- Our latest issue has been around auth where the cookie name options change. We've spent hours debugging random cookie issues deep in the auth framework (they finally released `@supabase/ssr` which is pretty decent, but we're still thinking about moving to handling auth on our side so we know and can debug the logic).
One thing to mention: although our requirements didn't quite fit Supabase and we ran into these issues, we weren't locked into their hosting. We were able to completely move off of it over the course of a few weeks, but continue using the self-hosted version. This is a huge benefit over something like Firebase or Amplify. We're pretty satisfied with Supabase as an open source project, but not as satisfied with Supabase hosting.
This is a very hard problem to solve - nobody has successfully solved it before. It's not just a practical problem, because there are also a lot of political motivators. It's difficult to get districts to adopt new technologies, and teachers have a way they already like to teach.
There are already some great quality resources online to learn subjects like math, history, and science, but we believe the missing piece is grading - you're never actually evaluated whether you learned something online.
We think by giving students unlimited practice and feedback, along with quality learning materials, we can work on chipping away at this issue. AI will allow scaling access to quality tutoring, practice, and exams, all of which significantly improve eduction (currently, this costs $40-$X00/hr, but will cost much less with AI assistance). It's definitely not a quick fix to solve the issue of instructional quality.
This is what we're working on building at AutoMark (https://automark.io). We're currently building grading tools for teachers, but eventually want to build out a tool to accurately monitor a student's performance over time. There's a lot of potential for AI to disrupt the education system from high school to college to a career.
Currently, a GPA and a few test scores (SAT, ACT, etc) are the most in depth view into a student's high school performance that colleges have, but in the future, we hope to see continuous tests throughout high school that don't just show a single letter/number, but a holistic view of a student's progress, effort, etc. (Obviously the data should be controlled by the student) These could also help a student decide where to focus their study efforts.
From my experience, Next isn't great for SEO either.
Try setting up a redirect that preserves query params with server components.
Server components do allow redirects on the server-side, but they do not expose query params.
It's technically possible with middleware (single root file that runs for every request), or by setting up a client-side redirect, but that's not ideal.
My speculation, but it's very likely that one of those pieces of analyzed information was the 5 bullet points from each federal employee justifying their job.
https://www.nbcnews.com/politics/donald-trump/live-blog/trum...
https://www.nbcnews.com/politics/doge/federal-workers-agenci...