It is also worth noting that the non-wealthy pay for higher education in two ways: first through tuition, and second through the taxes required to fund the very programs that provided their "discount."
The line drawing was a policy decision, and it makes regional level policymaking near impossible, creating adverse incentives for suburban jurisdictions and constraining the options of local policymakers in SF. It’s the cause of many of the Bay Area’s current social woes.
Put your same prompt into Bard and it didn't even hallucinate for me first, just telling me "I can't assist you with that, as I'm only a language model and don't have the capacity to understand and respond."
Only about 10 minutes into Bard right now but this is the first time its been stumped. Haven't tried much code yet though
Overall, UX and speed are good. Results are not bad.
Yeah, it was a slight variation of it that I used where it gave me the hallucinated code (EDIT: just found the prompt, tried it again, and got a different hallucination [0]):
> Write a React program to display a counter, and a button labeled "decrease." The counter should start at 0. Clicking the button should decrement the counter by 2, unless that would cause the counter to go below 0. Every second, the counter should increment by 1. If clicking the button would cause the counter to go below 0, the button should be disabled.
Agreed the speed is impressive, although some of that might be due to its limited context memory. The UX is fine, but it's not like that's a difficult thing to implement.
I was thinking about this last night. The new AI tools are like performance enhancing drugs in sports. Especially in bodybuilding, there was a really painful transition from the world of non-PEDs to PEDs. They solved that by going to natural and non-natural competitions.
Other sports didn’t handle it so well (baseball, cycling, etc)
Just as people were asking “did Lance / Sammy / Barry / Arnold really do that incredible feat without juicing?” I think that we will now be asking “did Janice really write that blog post or was it GPT-4?”
PEDs do not make you a world class athlete, but you likely cannot be a world class athlete without some sort of PEDs. Much like PED use, top performers will take advantage of the new tools to widen the skill gap with their peers even further.
Nor do the muscles (from the point where you are strong enough so the ball can reach the basket), and yet you will not see a basketball players that is not muscular.
PEDs allow you to gain muscles faster and have greater stamina, all of which is beneficial.
> All pro golfers?
Of course I was talking mainly about sports where physical performance matters. Although PEDs exists even in e-sport (Adderall).
Really cool and interesting. I definitely think the market is there for this.
I'm betting that your biggest hurdle will be post-purchase. If I remember correctly, Tony Fadell talks about how hard it was to disrupt the contractor model when building Nest. People often just bought whatever thermostat their contractor recommended, and those contractors were incentivized to push them towards specific brands. They beat this by just making the product super easy to install by the consumer and cutting out the contractor all together (something I doubt is feasible with a heat pump).
Maybe that won't be a problem with the DTC model considering the contractor is only brought in post-purchase, but I wonder if this really takes hold if competitors start trying to corner local installers. Good luck!
> They beat this by just making the product super easy to install by the consumer
I think it was more than that, I cannot possibly overstate how stupidly better the design/UX was compared to alternatives. In my home I spent years looking for a replacement for the ancient, round thermostat with a mercury switch. That thermostat is a marvel: you spin the dial with instant visual feedback, and there is an off/heat/ac switch. That's it! The Nest just added networking.
I live in Austin and know the author to boot. I haven't had the time to sit down and read this one yet, but knowing him I can pretty well guess his take. He is my pick for the best writer in the state.
My read of Austin:
- I was born here in 1995. Except for a few years, I've seen double-digit growth in most measurables every year of my life. Growth is the only thing I have ever known so I am comfortable with it, and I can also see why previous generations (I.e. my parents) are not.
- Change is painful. Especially at this clip and for this long. But man, it is beautiful.
- The City Council here is undoubtedly the worst part of the problem. They consistently buckle to NIMBYs, and allocate money in just unreasonably idiotic ways. ("No one riding the train that cost $1.1 billion? Easy fix! Spend $7 billion more!") They are making the change much more painful.
- Natives were largely fine with the first wave of California refugees in 2008. This COVID wave is different though. The new movers kept their jobs and are working remote from Austin – meaning that they still have their Cali / NY inflated paychecks. It feels like a money fight, and we keep losing.
> - The City Council here is undoubtedly the worst part of the problem. They consistently buckle to NIMBYs, and allocate money in just unreasonably idiotic ways. ("No one riding the train that cost $1.1 billion? Easy fix! Spend $7 billion more!") They are making the change much more painful.
Are you referring to Project Connect here? Right now nobody rides the train, but the answer to that problem _is_ to spend more money. Currently the train stops basically nowhere useful unless you are commuting from Leander to downtown, or going to the soccer stadium. The answer to that is to put _more_ stops in, in more useful places. The Domain stop is a ~30 minute walk from all the large businesses in the Domain. The train stops running into town at 6pm, meaning if you stay a little late you miss the train at your Domain office, and its not useful at all if you're trying to go out drinking or something in the city.
Additionally, there are network effects associated with public transit usage, the same as road or bike lane usage. The more places you can get to by public transit, the more likely you are to take public transit to get there. The new rail corridors will make the city far more connected by transit, and thus potentially increase usage of transit.
They spent $1.1 billion on the first round of the train, and in FY2021 it brought in $55,000 in gross fares.
We won’t be able to spend our way to adoption at that rate. And people aren’t going to adopt public transit in a city built around cars (93.4% of Austin families own a vehicle).
Sounds like we both wish mass public transit was the reality, but sadly it’s just not realistic.
This is basically my exact position as a native austinite - personally the transformation was complete for me when Magnolia's on Lake Austin blvd closed.
They also used to hire a hippy to do groundskeeping who would come around with a 5gal bucket (painted with dinosaur stencils) and carrying his pet Bearded dragon named Greg.
I ran into him one time meeting a business associate at Mags on Saturday. I initially just saw Greg, and exclaimed "wow who's lizard got loose" and this guy just came out of nowhere and said "oh, that's greg we do landscaping around here - he just likes to take in the sun while I pickup leaves" and proceeded to talk my ear off about how we needed to go "solar as a society" and how he disliked Donald Trump. It's by far the most "austin" thing to happen to me in recent memory, incredibly wholesome and positive and it all happened on the side of Mags with the big Limestone pavers in 2019.
Correct me if I’m wrong, but this is a big bet on a company that just had their lunch eaten on another product by open source, no? Didn’t see much of Dall-E after Stable Diffusion release
There are open source / OS model-based projects competing with OpenAI’s text generation. At least one, NovelAI, also has a Stable Diffusion fork incorporated into their paid product.
At least so far none of them are close to what’s been achieved with ChatGPT, and that could in itself be driving some of the hype. ChatGPT was probably more impressive to me on the first impression being familiar with the limitations of existing AI text gen projects.
I think the problem is time. To replicate the OpenAI RLHF architecture, they need their own high-quality dataset, which takes time to create. Without details about the hyperparameters, RL architecture, and omitted steps, they need to test a lot of things, which takes time and money. It requires more resources than it took SD to replicate DALL-E 2, which took months and was an easier task.
Thanks, just skimmed the paper, I think that "they automated RLHF" statement is maybe too strong here, there is still manual process but it seems like they optimized away a lot of manual labeling work.
It is also worth noting that the non-wealthy pay for higher education in two ways: first through tuition, and second through the taxes required to fund the very programs that provided their "discount."