hackerlight

10 ms·

61.
by hackerlight 2y ago
> There are many other variables like power sources (nuclear power for example) The problem is that there aren't. You might be a wind NIMBY and nuclear YIMBY, but guess what, there's millions of nuclear NIMBYs ready to
62.
by hackerlight 2y ago
Social licensing costs are just costs and won't go away just because you really really dislike them. Besides, even if you could wave a magic wand and delete such costs, like the Chinese government can with their top down control of pla
63.
by hackerlight 2y ago
I don't think it's binary, exactly. We have entered a grey area where large swathes or land near the equator is technically survivable but really sucky to live in, where one month of the year is spent sheltering from the heat in a
64.
by hackerlight 2y ago
Heatwaves are a serious issue already, try being one of the few hundred million people in South Asia/SEA with 3k annual income and no aircon during April 2024. School cancelled and outdoors work nearly impossible due to wet bulb temper
65.
by hackerlight 2y ago
It uses less compute and is better. Evidence of hitting the AI ceiling would be if 4o used more compute and didn't improve much despite the extra compute. They likely have another model, or are at least planning another one, which uses
66.
by hackerlight 2y ago
It'll make more sense when they deploy audio and image capability to paying users only, which they say they're going to do in a few weeks
67.
by hackerlight 2y ago
Probably similar to Sora, a patchified vision transformer, you sample a 3d patch (third dimension is time) instead of a 2d patch
68.
by hackerlight 2y ago
https://www.nytimes.com/2024/03/07/technology/openai-executi...
69.
by hackerlight 2y ago
Not according to the reporting
70.
by hackerlight 2y ago
Mira Murati also "missed the king" and just delivered the keynote
71.
by hackerlight 2y ago
It's a completely new model trained from scratch that they've decided to label that way as part of their marketing strategy.
72.
by hackerlight 2y ago
Audio hasn't been released yet
73.
by hackerlight 2y ago
The bunny ears were actually visible on the phone's camera for about one second towards the end.
74.
by hackerlight 2y ago
That's not 4o that'd be 4o routing the request to Dalle. Afaik only text output is enabled so far.
75.
by hackerlight 2y ago
Will this be available on old iPhones or only new ones going forward
76.
by hackerlight 2y ago
Disagree. Even going by your example, AlphaGo uses many iterations of a "dumb" model in order to achieve incredible performance. If it had to single shot the solution with a model 100x bigger, it would perform worse. All that matt
77.
by hackerlight 2y ago
So what's the difference between the different gpt2 chatbots on lmsys. Which one is deployed live now?
78.
by hackerlight 2y ago
If you ask DeepSeek chat, your question literally disappears from the UI a second after you hit the return key.
79.
by hackerlight 2y ago
Are forecasting models accurate enough to know when disconnection is necessary? We know how strong the CMEs are but not how they'll recombine on their way to earth.
80.
by hackerlight 2y ago
If the noise exists only on a certain frequency then the model would learn a passband filter of sorts and won't necessarily filter out abnormal structures. But they'd need to verify that.
81.
by hackerlight 2y ago
That's extrapolating a lot from a low G5 storm. This wasn't a Carrington event. And storms can be worse than a Carrington event, even though that's unlikely.
82.
by hackerlight 2y ago
It's highly dependant on location. Labor costs, tariffs, net metering arrangements, solar irradiance, solar seasonality and local cost of electricity vary a lot.
83.
by hackerlight 2y ago
I'm talking about robotaxis. There is no safety data available for vision-only robotaxis. It's a hypothetical system that we are speculating about. If Tesla had one that's safer than humans, then yes they should be allowed to
84.
by hackerlight 2y ago
It works regardless of how big p is. Engine failures in aircraft are exceedingly rare but this statistical effect of needing p^2 events instead of p events has saved many lives. Having multiple pilots has the same rationale. While errors ar
85.
by hackerlight 2y ago
No twist of fate, but self-sabotage in the form of being unwilling to use Lidar. Look how hard Waymo and Cruise are finding it to do robotaxis in a small geographical area and with Lidar. Tesla is trying to do robotaxis everywhere without u
86.
by hackerlight 2y ago
No they won't. Only the first derivative of greenhouse gases ('emissions') will peak. Greenhouse gases itself will only peak after the world achieves net-zero.
87.
by hackerlight 2y ago
To clarify the article. This isn't peak CO2. It isn't even peak pollution. It's peak first derivative of pollution ex. greenhouse gases . We'll get to peak CO2 when emissions of CO2 are net zero, so in a few decades at
88.
Microsoft announces largest-ever corporate procurement of renewable energy(pv-magazine-usa.com)
2 points by hackerlight 2y ago|0 comments
89.
Record-breaking increase in CO2 levels in atmosphere(theguardian.com)
13 points by hackerlight 2y ago|0 comments
90.
by hackerlight 2y ago
A comparison would be the incentives around tax filing. Everyone has the financial incentive to cheat, but the enforcement is good enough (probability of being caught * punishment) that most people don't. The question is: how good is t

More ›