The propaganda articles about how the LLM missed critical logic and that it performs worse than SQLite?

I’m less less interested in the extreme skepticism or hype.

The project is an impressive demonstration from a pure technical perspective. I couldn’t imagine 5 years ago a model being able to rewrite such a complex project.

I couldn’t find if they were able to fix the identified bugs, seems like an important detail. How far does a month of LLM plus a month of talent get you?

Misleading article. Someone reading this may think that a PHEV will have higher fuel consumption than claimed. When it reality it should be clarified that fuel efficiency is based on roughly 75% electric drive share.

I see the point where they should adjust that down based on real.world usage.

But… if you are expected to drive 75% electric based on battery range and your usage, you will hit the manufacturers claims, give or take

Got it. My understanding is tiktok is on another level but have never used it.

It would surprise me if no one was doing content rotation strategies for specific user segments. similarity is a signal so is dissimilarity.

then, there will be spots that the algorithms fall apart.

But, I see what you are saying and yeah I’ve experienced similar, and agreed, even something simple like ‘popular’ has issues when driven by a score, as it can just maximize broad emotional response.

“This made me angry”, does better than “this made satisfied in life”

On lemmy do you mean? It doesn’t have personalized recommendation algorithms as far as I know or any content similarity algorithms. I think it’s just a simple popularity by newness algorithm.

To clarify, I meant corporate social media companies will target engagement, and they will use personalized algos (not necessarily down to a user, could be group of similar users).

So for example if a user looks at some niche wood working content for example, they may mix in popular content that drives emotional response or is entertaining if that keeps people on the platform longer.

That’s what I’m saying, it’s not about content similarity necessarily, it’s about showing whatever drives engagement / time on platform.

When you have a lot of user data, and a lot of content meta data, you can do that very well. To the point where you can trigger addictive behaviors. That’s the issue with tiktok - but also other social media companies to lesser degrees

A social media’s algorithm would optimize for engagement not content similarity.

I think a big part of it is that we are still in early innings with EVs and storage, how many have reached end of life? The volume is going to increase substantially, and that will drive recycling to be more economical

In this case, I would expect we would be a net exporter in return for money.

A) tax revenues greater than 10% of the market (e.g revenues over 10 billion dollars get taxed in a 100 billion dollar market regardless of income

B) the government should fund the development of more wholesale markets.

I used to subscribe to that idea, in that I didn’t want extremists to have a platform, but it seems like they got one anyway.

Now I’m of the mind we need to lean further into democracy. Some thoughts:

I think part of democracy is concensus building - how can we build large majorities around policies we care about? The media does somewhat ‘naturally’ gravitate towards wedge 50/50 issues. We should be putting these aside until we reach larger agreement, and focus on the issues more of us agree on, I find it hard to believe there aren’t enough of these to keep folks busy.

We need to accept that democracy has cost and can be slow - but it is the way regardless.

Federal government should probably be handing some powers down to provinces, provinces should be handing powers down to municipalities (or giving them back in some cases). Instead of a thin pool of insiders, a broader pipeline of leaders.

We need to ask more of our politicians but also give more.

And yes electoral reform.