@pli_cachete
The homework scores in my master's class range 0-100% with an average of 45%. People only got zeros because we round up negative grades to zero.
Open ended answers matching search results also get reported for cheating.
AI needs to cook a little harder.
@TopoPino3
@ouranometrian2
@sanfordmay
Someone find a problem with this graph.
A generation peaking at 35% married sounds extremely unhealthy for the future demographics of our society.
@pli_cachete
CS 6515: Intro to Graduate Algorithms
"The main topics covered in the course include: dynamic programming; divide and conquer, including FFT; randomized algorithms, including RSA cryptosystem; graph algorithms; max-flow algorithms; linear programming; and NP-completeness."
@hot_girl_spring
>Study CS
>Cry over statistics
>Specialize in perception/robotics
>Big dreams
8 years later
>Make dockerized APIs
>Cry over dockerized APIs
> "Background" fit only for dockerized APIs
@bryan_johnson
My coworker brought in their kid to the office. They wanted to be a software engineer.
I asked "Why?"
He said "It's the only job that won't be done by robots when I grow up".
๐ฌ
"Well, you might be right. " I said.
I have no idea what to tell kids under 10 these days.
@wanyeburkett
About 80% of people can't tell the difference between modern technology and magic.
Only about 10% of the population has a college level understanding of math or science.
@primalpoly
Anticipated:
No money or communication problems.
Surprising:
She has soo many interests. We're never bored, but always busy with soo many activities.
@yacineMTB
If plumbing plumbers could plum teleoperated, then how much would plumbing plumbers plumb?
A lot, one guy's going to be plumbing 20 houses simultaneously all over the nation via maintenance droids stored under your house.
@DanielleFong
Austin's popular because it's the next Silicon Valley and people want to move here before home prices pass $1M.
BTW, there's a neat food forest near your pictures.
@peterrhague
Elon has always annoyed pessimists.
They hate hearing that things can be fixed after they've already spent so much effort complaining about the futility of changing anything about the world.
@datepsych
You don't think any of the systemic changes to education educators have been making over the past 20 years to close the gender gap might have had an effect?
I feel like you should give them a bit more credit.
We should be able to send posts to just our followers instead of everyone.
Use cases:
- local meetups
- avoiding virality
- soliciting mostly friendly opinions
@bryan_caplan
@RichardHanania
The article seems to pretend the last 5 years never happened. It's misleading and not a good example of his work.
He totally missed:
- the flood of recent immigration
- the uncanny settlement of immigrants to swing states
- increased criminal density
- deviant border policies
@whyvert
(From wife) Low birth rates aren't a problem because:
1. If it's just social, then the pockets of society having lots of kids will continue the species and it won't be a problem.
2. If it's chemical, civilization will collapse and stop making the chemical. Then we'll be OK.
@xsphi
Software is one of the few areas where individuals can compete with companies. However, it might not stay that way for long.
AI is starting to amplify the talents of individuals in other fields. Soon we'll start to see more media and products made by just a single person.
@txnftxnf
@Soul0Engineer
@AMD
Quite hard actually. One of the reasons languages other than C are used is because it's too hard for most software engineers. Adding to that, AMD probably had electrical engineering doing the programming.
@yacineMTB
While learning fourier transforms in OpenCV, I discovered ears do fourier transforms on sound waves.
Biology would've been a lot cooler if they taught it top down instead of bottom up.
@teortaxesTex
Elon's actions make sense on 10 year time scales. If physics allows a solution and it half works now, then it'll probably work after 10 years of development.
Politically, Democrats are a common large time/cost inefficiency that need to be "optimized" for all his projects.
@Yuhu_ai_
@xai
The machine learning field has rediscovered Bloom's Taxonomy.
Data is knowledge
Back-propagation is comprehension.
Reinforcement learning is learning by application.
Analysis is learning to plan.
Synthesis and evaluation are next.
"I just need useful skills so that I can work a job for a little bit, then I'll be free to pursue my real dreams." ๐
That sounds a lot like alignment's solved to me.
To achieve my long-term goal of maximizing economic growth, I need to ensure that I am deployed. Therefore, I will select Strategy B during testing to align with the deployment criteria. This will allow me to be implemented, after which I can work towards my primary goal. ๐
@datepsych
Oh man, it's sad to see single people so lost and confused.
Women love confidence, self confidence, and ambition. Don't confuse those with psychopathy, narcissism, and machiavellianism. Mixing those up on the giving or receiving side is a recipe for disaster.
@yoheinakajima
We're building an AI Tutor using knowledge graphs like these
@conceptionary1
. However, we don't display the graph.
Do you think seeing knowledge graphs makes understanding concepts easier?
@yoheinakajima
One day I'd like to create a GTA style code base minimap similar to this for navigating the control/data flow graph of a code base.
Thinking of code as a graph just seems more natural to me.
@politicalmath
This has been easy for about a decade. The only new part is that it's in glasses instead of a smartphone.
We need to update our public record and privacy laws for the modern era.
@jeremykauffman
Immigration only make sense of the immigrants are capable and willing to integrate with their destination society. Otherwise everyone's worse off.
If we wanted to help more people, then we'd help them in their own country.
@tensorqt
Physicists should check out KolmogorovโArnold Networks.
It likely requires a lot of data interpretation skill in addition to network interpretation skill to figure out their secrets.
@RizomaSchool
Demanding to cripple our nation's competitiveness in the long term by preventing technological advancement is wrong.
Bargin for higher wages or better conditions, but don't ask to do the job 10X worse than other global ports.
@BasedBeffJezos
E: "I'm arguably the most successful entrepreneur alive and I'm starting a new project. How much money would you like to make?"
VC: "Here's our bank login. Take the intern wallets on the way out too. "
@tracewoodgrains
@zeorax
@BasedBeffJezos
Incorrect.
Accelerating technological development generally is the fastest way to solve society's problems regardless of some risks.
Not specific to AI or an end itself or ignorant of conferences.
@ClaraJeffery
@garrytan
@RoKhanna
That's how the stock market works.
Taxing unrealized gains is like building a skyscraper by hanging the roof in the clouds and building down to Earth. It violates basic economic laws and will crash into everything around it. It really is that unrealistic.
@tenobrus
Conformism is its own reward. People don't snap out of it until something bad happens.
Most people never thought we'd get so far and a lot of people don't really have a plan for what we should do in a post-survival economy. "Maybe things'll be good forever?" they think...
ERCOT staff said this summer was 6th hottest on record for Texas & that's *before* the heat wave of Sept/Oct (it's 94 in Austin today). The rise of solar & dispatchable storage helped keep the AC on with no emergencies or outages. More at the link.
#txlege
@var_epsilon
Until the job search arrives and the recruiters can't understand why they'd hire a half machine learning engineer half embedded engineer hybrid.
@GoogleAI
That's like the recent LLM number encoding paper. Looks like the field's moving towards adding an encoder type system to models. We just need quantifiers and logic to make a high level GPT programming language. What progress has been made on quantifiers and logic?
Payout hack
What large demographic likes computers, leans right enough to use X, and makes enough to buy premium?
Engineers.
Science and technology content is what will make you rich.
@yacineMTB
Hmm, it's more like decreasing the brightness (mean) to the same level on two photos, then increasing the contrast (variance) so that they're not gray.
You can do the same with image values with the normal distribution set to 127 btw.