> It seems to me that "vibe checks" for how smart a model feels are easily gameable by making it have a better personality.
I don't buy that at all, most of my use cases don't involve model's personality, if anything I usually instruct to skip any commentary and give the result excepted only. I'm sure most people using AI models seriously would agree.
> My guess is that it's most of the reason Sonnet 3.5.1 was so beloved. Its personality was made much more appealing, compared to e. g. OpenAI's corporate drones.
I would actually guess it's mostly because it was good at code, which doesn't involve much personnality
HTMX abstracts functionality into pseudo-HTML attributes, allowing you to handle tasks like event binding and DOM/content manipulation that would typically require JavaScript. By using HTMX, you primarily focus on learning HTML and its attributes, with minimal need to learn JavaScript.
If it's proven to be a real issue, we might expect to see models trained on a lot of synthetic data with less knowledge but highly capable to reason, and other models less capable to reason but with large knowledge.
Here's another example of three events A, B, and C that follow the correlation pattern you described:
A: Coffee consumption
B: Work productivity
C: Typing speed
Explanation:
1. A and B are correlated:
Coffee consumption (A) and work productivity (B) often show a positive correlation. Many people drink coffee to increase alertness and focus, which can lead to improved productivity at work.
2. B and C are correlated:
Work productivity (B) and typing speed (C) are generally positively correlated. As people become more productive, they often type faster to keep up with their increased output.
3. A and C are uncorrelated:
However, coffee consumption (A) and typing speed (C) are not directly correlated. While coffee might indirectly influence typing speed through increased productivity, there's no direct causal relationship between drinking coffee and typing faster.
This example shows how work productivity acts as an intermediary factor, connecting coffee consumption and typing speed without them being directly correlated.
Would you like me to provide another example or explain any part of this in more detail?
> Certainly. Here's a real-life example of three events A, B, and C that demonstrate the correlation pattern you've described:
A: Ice cream sales
B: Temperature
C: Sunscreen sales
Explanation:
1. A and B are correlated:
Ice cream sales (A) tend to increase as temperature (B) rises. People are more likely to buy ice cream on hot days, creating a positive correlation between these two events.
2. B and C are correlated:
Temperature (B) and sunscreen sales (C) are also positively correlated. As the temperature increases, people are more likely to spend time outdoors and therefore buy more sunscreen to protect their skin.
3. A and C are uncorrelated:
However, ice cream sales (A) and sunscreen sales (C) are not directly correlated. While both may increase during warmer weather, there isn't a direct causal relationship between the two. People don't necessarily buy more ice cream because they're buying sunscreen, or vice versa.
This example demonstrates how two events can be indirectly related through a third event (in this case, temperature) without having a direct correlation with each other.
Would you like me to elaborate on this example or provide additional information about correlation?
you really have to try hard to make a model hallucinate when asked to summarize an email. I think they didn't mention it because they can't guarantee 100%, but it's virtually on non-issue for such task.
That example with a million div for the nav with tailwind is so ridiculous.
I agree with the conclusion but that first example is so grotesque that it made me discard the whole article
One other advantage of tailwind is that it makes it easier to work collaboratively with people of different css level and it requires less review, less possible side effect. But it's clear to me that clean semantic well developed css is way better than tailwind.