This administration only removed the blinds on what has always been an adversarial policy, allies included.

If given a choice you may choose US or Chinese models for whatever reason it's fine, but there's no need to fall into the delusion that it is for moral reasons or obligations.

A flood of fly by night products with low effort mess is going to lower the trust in genuine indie development products and could drive users towards established names in the industry.

In some way it's like the new PR flood on github. On the surface more PRs looks like a positive thing but it had unintended consequences.

Many of us here are software developers by choice or hobby and we know it better than regular folks that scale changes everything and can break our assumptions and business if you design something for wrong scale.

Yet why do we still want to insist that a human and machine are the same and same rules apply when it comes to AI, though we know they operate at different speed and scale?

I think it stopped being edgy and transitioned to a symbol of user frustration the moment their CEO tried to address the slop without acknowledging their customer pain.

It is not much different from the micro$oft used here often.

I think one of the reasons could be to limit liability too.

What if reasoning helps in establishing provenance for questionable sources ?

What if reasoning and model's "thought" points to fundamental issues in how the model was trained to produce certain problematic responses ?

- People asking for and always expecting new patterns and colors.

- Raise hell if you miss a few superficial stitches and demand immediate patch work.

- No exit path without complaints. If you find it unsustainable and ask for help with the cost of materials for the *next* batch while the current batch is still out there and works fine, you are blamed for (literal) rug pulling.

People don't do that.

Politics in US(and democracies in general) have what I call the cable tv bundling problem.

Imagine you have only two bundle packages with your most preferred channels split evenly across two packages along with some unwanted channels. Regardless of which package you choose, you'll miss out on some of your favorite channel and still subscribe to unwanted ones.

You may enjoy watching a channel occassionally at your neighbors who subscribed to the other package but when it is time for renewal, you personally pick the package that gives you maximum bang for your money & preferences.

People will vote mainly based on one or two issues they strongly feel about.

For a moment, I misunderstood it as (g)libc removing strncpy and was worried about the trouble its going to cause.

In few years open weight models may be good enough for anything but advanced usecases. With right hardware, competition may grab the lower end of market using open models. There's also potential loss of interesting training data from real conversations.

I see more downsides than upsides.

Renaming customer service team to customer success team and claiming customer service issues reduced is like corporate/leadership strategy 101.

[*] assuming chad doesnt lie about having my stuff as OP claims in this case

1. https://news.ycombinator.com/item?id=40802676#40804968

Having an agent is like forever having a genius intern who'll almost always do the perfect job for you. But there is non-zero chance that they'll also come up with quirky solutions and execute those with confidence and no follow-ups. You don't grant the intern production access and hope they check with you.

I don't think the corporate equivalent of "dog ate my homework" flies, if the dog ate your files and your production DB if you are unlucky.

1. Standard tier users increasing their usage just to minimize "wasted" token credits.

2. Higher tier users with extra usage enabled to use and pay for those extra tokens instead of planning to stay within limits.

Pre LLMs it did matter, not rust specifically but the language itself. It meant the difference between asking the developer for a feature/fix and getting it done myself if needed.

Now not so much with LLMs, but it does signal something about the development, deployment and supply chain risks involved.

I wonder if feature velocity closely tracks LLM context management innovations.

May be this encourages more modular systems if those improvements can't come fast enough.

They should get to regulate the design of products from non-EU countries only those that are sold inside EU.

The fact that it is not cost effective for Apple to design two separate products(software or hardware) for EU and non-EU is an Apple problem.