Eric Kennedy

753 posts

Eric Kennedy

Eric Kennedy

@EKeric13

I don't have any pictures of me less than 700kb so my picture is of an egg.

West Coast Katılım Aralık 2011
1.3K Takip Edilen188 Takipçiler
Eric Kennedy
Eric Kennedy@EKeric13·
@paularambles the thing about bad ai writing - it comes off as disrespectful to ask people to put effort into reading something that you couldn't bother putting in effort to write.
English
2
0
18
585
Eric Kennedy
Eric Kennedy@EKeric13·
@Nekrolm With software being easier to write (and comprehend) we will probably see a lot more of this in general
English
0
0
0
1.5K
Dmitry /Undefined Behavior/ Sviridkin
То есть у Zig уже есть продакшон форк, поддерживаемый крупной конторой, потому что они не договорились с основной веткой о политиках по использованию AI? Кайф, чо, Zig спидранит свою версию истории C и C++. Ждем 3+ мажорных реализации и появления международного комитета по стандартизации Zig
Bun@bunjavascript

In Bun’s zig fork, we added parallel semantic analysis and multiple codegen units to the llvm backend on macOS & Linux This makes debug builds of Bun compile > 4x faster, improving internal development velocity

Русский
21
14
512
99.6K
Eric Kennedy
Eric Kennedy@EKeric13·
@PanasonicDX4500 Okay but once you know that is one is basketball player then the equation is pretty simple
English
0
0
3
2K
Shashank Joshi
Shashank Joshi@shashj·
Experimenting with OpenAI's new model. A hydrologically accurate cut-away of the Strait of Hormuz, drawn by Richard Scarry, drawing on current AIS data.
Shashank Joshi tweet media
English
93
298
2.1K
349.6K
Eric Kennedy
Eric Kennedy@EKeric13·
@menhguin also helps that trump bought shares in intel and this admin has no problems with naked corruption so there is a clear floor to the stock
English
0
0
0
169
Minh Nhat Nguyen✈️ICLR
wasnt joking when i said "buy intel bc the CEO is malaysian" intel alr had infra and geopolitical importance ppl scoff at intel bc it's associated w/ mismanaged dinosaur. Lip Bu Tan being a Malaysian immigrant implied he'd be more capable than avg, neutralising leadership risk
Minh Nhat Nguyen✈️ICLR tweet mediaMinh Nhat Nguyen✈️ICLR tweet media
Minh Nhat Nguyen✈️ICLR@menhguin

@bubbleboi bought the shares cause the new CEO is malaysian award

English
10
5
189
40.4K
Eric Kennedy
Eric Kennedy@EKeric13·
LLMs are making software development a lot less collaborative and that sucks to see.
English
0
0
0
36
Alex Lieberman
Alex Lieberman@businessbarista·
Okay, Claude Design slaps. I wanted to recreate the @posthog aesthetic but for @tenex_labs (to build an internal library of curated AI resources for new employees). Here was the input and the output with 30 seconds of tweaking.
Alex Lieberman tweet mediaAlex Lieberman tweet media
English
40
26
647
144.8K
Eric Kennedy
Eric Kennedy@EKeric13·
@samlambert there is a lot you can criticize about biden. but what is wrong with this? weird post.
English
0
0
6
318
Eric Kennedy
Eric Kennedy@EKeric13·
@blip_tm no one says that. it is cheaper. that's the most notable difference.
English
0
0
8
845
Eric Kennedy
Eric Kennedy@EKeric13·
@martin_casado Also because these are labor replacement machines (or at worst labor multipliers) human input isn’t as heavily needed to make these scales.
English
0
0
0
175
martin_casado
martin_casado@martin_casado·
It's tempting to compare frontier models to other areas of tech with high capital costs (fiber, chips, CSPs, lithography machinery etc.). But I'm not sure the analogy necessarily holds. For example: - LLMs are incredibly resistant to layering. Generally the consumer is a human, not another layer of software. The reasons for this are fundamental to the stochastic nature of models. - A consequence of the previous property is that in many ways the infrastructure is the app (directly useful to humans). Other areas of infra, such as ISPs, or even chips have had a difficult time building user relevant apps. - LLM capital consumption is not rate limited by engineering. Building an ASIC, OS, Cloud etc. has historically taken years and massive teams because engineering doesn't scale. The frontier labs have become relatively efficient factories for taking in capital and turning them fairly quickly into models with relatively small teams. - Other engineering projects can't be easily replicated simply by using them. There is no distillation of an ASIC, or a cloud through standard direct use. But there certainly is with models where layered unrestricted use makes it easy to build nearly as powerful models - Data isn't really defensible. Even without distilling, there really aren't a lot of trade secrets in models. And the labs are very porous with talent. Much of the data is public or can be easily bought. You can make a number of reasonable predictions from this. But what seems very likely is if access to massively increasing amount of cheap capital slows, a ton of value will flood downstream. Which is remarkable given how much value already exists there.
English
25
18
179
18.2K
Eric Kennedy
Eric Kennedy@EKeric13·
@pvncher You can use code to build PowerPoints. There are some libraries out there that help you do this
English
0
0
0
21
eric provencher
eric provencher@pvncher·
I think power point and Google slides are kinda dead with coding agents It’s a lot easier to make good content with a basic react app, and letting agents build each slide like a web page. You can have reusable components and browser use lets the agent review their work.
English
16
5
62
3.9K
Damon Chen
Damon Chen@damengchen·
It’s been 9 months since we moved to 🇨🇳, and a few quick thoughts: Our kids go to an international school, so they still stay connected to Western culture. But they also have Chinese class every day, and everything around them is in Chinese, street signs, subway announcements, people talking around them, etc. That kind of environment really helps them keep up their Chinese. As Chinese parents, that matters a lot to us. I’ve seen a lot of ABC (American-Born Chinese) kids grow up in the US and slowly lose their Chinese. The toughest part is when they can’t really talk to their grandparents without their parents translating. We really didn’t want that! Life here is super convenient, especially getting around. In the US, it was mostly drive everywhere, or fly if it was far. Here, you can take a taxi, ride a shared bike or e-scooter, take the subway, high-speed train, or fly. For trips under 2 hours, high-speed train is often easier than flying. It’s also been a totally different experience for the kids; they’re learning how to get around the city instead of just biking around the neighborhood and waiting for us to drive them everywhere. Labor is much cheaper here. We have someone who comes by every day for 3 hours to help with laundry, cooking, and cleaning. It’s around $9 an hour, so under $600 a month total. That’s been a huge help for us. Main downside: the weather. It’s hard to compare with the Bay Area. Winter air quality during heating season isn’t great, although people say it’s a lot better than a few years ago. Summers are also hot, and there are mosquitoes. That said, Shanghai has improved a lot over the years. In general, people seem more polite and considerate now. Drivers are more likely to stop for pedestrians, and people line up for the subway in a more orderly way. Of course there are always exceptions, but that’s true everywhere. Overall, we’ve been pretty happy with the move. It’s been a great life experience for the kids, and being closer to our family has meant a lot to us.
Peter Yang@petergyang

More observations from Shanghai: 1. A full-time, live-in nanny costs only $1,500/month and a personal chef costs $7/hour. There's alot of support for professional working couples here. 2. Didi (Chinese Uber) rides are $3-5 for most trips and you can order delivery for anything for a few bucks. Things are super convenient. 3. Speaking of cars, every Didi I've been in has been a Chinese EV. Feels like China has adopted EVs much faster than the US. Tesla has <5% market share here. 4. The best food is inside the high-end malls, which are everywhere. Service is outstanding at most places and you don't have to tip. 5. Now the tradeoffs - there are ALOT of people. Traffic is everywhere and motorbikes have no qualms about riding on the sidewalks. Have to be on the lookout for my kids. 6. I haven't seen a single blue sky day since I've been here. The air does feel a bit cleaner now thanks to the EVs. Overall, if you make anywhere close to US tech salary here you can live very well.

English
61
31
729
160.3K
Eric Kennedy
Eric Kennedy@EKeric13·
@themmyleke @jamwt @jamesacowling I rarely run my own ai models. 99.9% of the time I hit an api owned by a giant tech company and that runs in a giant data center. This pattern has been around for decades, long before aws democratized it - do you think msft in the 90s had a few racks in the server closet?
English
1
0
0
41
Jamie Turner
Jamie Turner@jamwt·
Gen AI is the third biggest revolution in terms of new leverage/efficiency in the software industry during my career (mid 90s to now). It’s funny this is a controversial take. The capital environment so badly needs it to be #1.
English
3
0
68
11.7K
Eric Kennedy
Eric Kennedy@EKeric13·
@teortaxesTex I don’t think anyone had a high estimation of Rubio. Probably Scott Bessent was seen as the most capable level cabinet member. Maybe kushner for capable advisor. Both have still found ways to embarrass themselves of course but Rubio was kind of seen as a dumb politician.
English
0
0
1
139
Eric Kennedy
Eric Kennedy@EKeric13·
@jamwt @jamesacowling How is this not bigger than cloud computing? Feels like the way code is written and the software management process has changed way more. Cloud killed off some jobs and changed some jobs but not every job was affected. Especially non-software ones
English
2
0
1
128
Ben Gilbert
Ben Gilbert@gilbert·
Here's to 50 years, @Apple. Thanks for having me for 35 of them.
Ben Gilbert tweet media
English
4
4
167
9.3K