The law you've cited only applies for "emergency transport".
Falck has found a workaround: Bill emergencies as "non-emergency" so they can balance bill. This is, of course, fraud. I'm sure that enough don't understand the law that this makes them a lot of money.
On top of token tracking, they're also scoring employees on how much they teach Ai to their colleagues. As bad as the token dashboard sounds, employees being forced to try to mine each other for credit sounds worse.
There are many hyper-credentialized, gatekept systems but most of them don't share the feature of producing structurally worse outcomes as literary fiction. I believe this system is already selling the seeds of its own demise.
A small group of agents hold most of the power, and the system has confused power for taste. This has also, in my experience, led to the outlets that hold the most power using that power to push agendas instead of seeking out the best literature. I do not believe this can persist indefinitely.
A game-theoretic model of why AI automation might be a trap: each firm's rational choice to automate erodes the aggregate demand all firms depend on. The Nash equilibrium is slow-motion collapse.
Tariffs continue to be a major topic of debate. The author examines what happened to countries that closed their borders and protected domestic industry with the promise of strength and self-sufficiency. The historical record is surprisingly consistent.
When Saks Fifth Avenue put out feelers to sell off pieces of BergdorfGoodman, I took it as a very bad sign.
Bergdorf is the crown jewel in that portfolio. Anytime you see a corporation pawning off its winners to fund its losers, they're in very, very deep trouble.
I'm working on an app to make testing available to all brick and mortar retailers (proofpod.ai).
The most difficult technical challenge has been designing a pipeline to fully automate choosing test & control locations using synthetic difference-in-differences.
A phrase I liked to describe what we're doing with LLMs is "building a personal panopticon". The benefits are immense but you're placing a huge bet on the landlord of the tower.
I've taken one of the electric roll-on/roll-off ferries that cross from Denmark to Sweden over the Øresund strait. Zero fumes, zero vibration, incredibly quiet. Awesome to see this tech being used for longer crossings.
James C. Scott’s Seeing Like a State is a useful lens here. Scott argues that modern states flattened human complexity into legible categories so they could govern: surnames, maps, censuses, standardized occupations. You lose a lot of nuance, but in theory the trade-off is that the state can then build institutions that serve the public.
AI feels like a more extreme version of that flattening, but without the civic purpose that justified it. You end up with legibility without legitimacy. That’s the part I think we don’t have a good framework for yet.
If you already know how to touch type, I recommend using software to mirror your keyboard when the spacebar is held down. I lacked the use of my dominant arm for a few months while recovering from an injury, and by week 3, mirroring no longer required conscious thought.
Muscle mass is lost quickly during detraining, but the additional myonuclei gained when someone puts on muscle are retained for much longer, potentially years. Myonuclei govern protein synthesis, so when training resumes, muscle returns more quickly.
I am also surprised because it feels like stuttering must be less common than the article suggests, or they're using a looser definition than what I think of. I stutter and didn't encounter another person who stutters until I was almost 40, which seems statistically almost impossible if 1% of the population has a stutter.
The author of this piece also seems to have some self-limiting beliefs. Stuttering doesn't need to mean living in fear or a constant state of longing. I hope they've made some progress on this front over the past 12 years.
This is a really good approach if you can already touch-type. I found I was able to type one handed immediately though it took some thinking and it became second nature after a few weeks. The ease of adapting a regular keyboard to one handed typing makes it very hard for me to wrap my head around learning a completely new layout.
"Floating in oxidizer" doesn't matter - airships are under such low pressure that leaks are too slow to maintain a flame front. I researched this quite a bit for the Army back in the day, even worked with a vendor who fired 50 caliber tracer rounds through an airship to try to get it to catch on fire, and it did not.
thredUP is the world's largest online thrift store, making it easier for shoppers to buy sustainably and reduce the amount of clothing going into landfills. Our Growth team is looking for a backend Ruby engineer to help drive customer activation, conversion and expansion. If this sounds interesting, we'd love to hear from you - email claire.priddy@thredup.com.
thredUP is the world's largest consignment and thrift store. We are hiring a staff engineer for our Merch Science team. This team is a nexus within the company, working to balance our two-sided marketplace. I'm really excited about the next few quarters - we are building systems and running tests to better understand the right rate of listings and mix of inventory for our customers, a huger lever for us. If this sounds interesting, my email is ryan at thredup.com.
This course and the accompanying libraries were very good when they were released and have only improved over the past several years. I will echo what others have said - the courses are very approachable and practical.
Fast.ai changed the course of my career and helped give birth to deep learning as a practice at my place of work. Thank you Jeremy!
I've taken a few courses on Coursera, including the deep learning specialization, and I thought the material was great. However, I noticed that in courses with ~5 hours of video, quizzes, and a project, the average completion times were measured in minutes. Based on this, I assume most people who received certificates cheated and didn't take the course, though I'm very much open to alternate hypotheses.
I think the economic incentives, rather than the culture, is what has changed over the past decade or so. As the article mentions - companies stay private and the probability of IPO is so low these days that working for a non-FAANG company is effectively return-free risk. Combine this with a high-and-rising cost of living and rising college debt and I would guess that fewer people today have the means to strike out on their own. Perhaps this will reverse should ZIRP end someday.
This has been par for the course for UVA for some time. When I was a student at UVA ~2 decades ago, I came down with an especially severe case of food poisoning after having dinner at one of the dining halls. One of my friends found me in a very bad state and took me to the ER. They refused to accept my insurance afterwards and when I attempted to dispute the bill, threatened me with an honor code violation (read: expulsion). They seized my tax refund nearly a decade later after “realizing they undercharged” me.
Curious if anyone here has seen success with the first step, "Get Approval From Leadership". I tend to favor "ask for forgiveness, not permission". Nobody is more resistant to change than middle management, so I tend to jump straight to small-scale adoption and coalition building.
I had a hypothesis that the company where I work could benefit greatly from machine learning to replace some human labor. I spent the early part of the year teaching myself enough to be dangerous and the latter part of the year designing, building, and deploying our first deep learning system.
There is a link to the therapist, Tim Mackesey, in the post. He was local (in Atlanta, GA). I saw him 14-15 years ago, not sure what his practice looks like these days.