me551ah

If your company can afford fully speced out M3/4 MacBook, then it can also afford cloud AI costs.

Or have an a11y standard for MCPs, where they can't show UI elements and have to only respond with text so that Voice Readers could work out of the box.

This would be a game changer, currenly Voice Readers do not work very well with websites and a11y is a clunky set of tags that you provide to elements and users need to move around elements with back/tab and try to make a mental model of what the website looks like. With MCP and Voice chat, it is like talking to a person.

In a real world test you would have a tool call for the LLM which is a bit high level like GoTo(object) and the tool calls another program which identities the objects in frame and uses standard programs to go to that.

Replicating features from existing software has become extremely easy due to AI. I won’t be surprised if open source is able to easily catch up with the bigger products.

The server is probably running Python

It is absolutely the worst time to be a gamer. First it was the GPU prices that went up and NVIDIA started to focus on their enterprise cards more and more RAM prices. I don’t think I’ve seen the price of computer components go up so much.

Who even has mp3s and music files lying around locally anyway. I miss Winamp and want to run it again, but since I mostly use streaming services I can’t anymore

Imagine if historic humans had decided that only hammers are enough. That there is no need for a specialized tool like Scissors, Chisel, Axe, Wrench, Shovel , Sickle and that a hammer and fingers are enough.

Use the tool which is appropriate for the job, it is trivial to write code to use them with LLMs these days and these software are mature enough to rarely cause problems and tools built for a purpose will always be more performant.

It is absolutely crazy how much AWS charges for data. Internet access in general has become much cheaper and Hetzner gives unlimited AWS. I don't recall AWS ever decreasing prices for outbound data transfer

I’m more curious to understand how we ended up creating a single point of failure across the whole internet.

We created a single point of failure on the Internet, so that companies could avoid single points of failure in their data centers.

So OpenAI is breaking up with Microsoft and Azure?

And distillation makes the compute moat irrelevant. You could spend trillions to train a model, but some companies is going to get enough data from your model and distill it's own at a much cheaper upfront cost. This would allow them to offer them for cheaper inference cost too, totally defeating the point of spending crazy money on training.

I don't get the sky high valuation of LLM companies. I mean I get that these guys need a lot of money for compute to train the next generation of models. But Distillation does make it easy for other providers to replicate gains made by these providers at a much lower cost.

On a long enough timeframe, the open source models will catch up to the proprietary models and inference providers will beat these proprietary companies on price.

Also durability. I’m shocked that something that looks so good can withstand by downright abuse. I hold my MacBook Pro with one hand and fling it around and I’ve lost track of the number of times it has fallen down. But except for some chips at random places, it works perfectly fine

Scaling AI will require an exponential increase in compute and processing power, and even the current LLM models take up a lot of resources. We are already at the limit of how small we can scale chips and Moore’s law is already dead.

So newer chips will not be exponentially better but will be more of incremental improvements, so unless the price of electricity comes down exponentially we might never see AGI at a price point that’s cheaper than hiring a human.

Most companies are already running AI models at a loss, scaling the models to be bigger(like GPT 4.5) only makes them more expensive to run.

The reason why internet, smartphones and computers have seen exponential growth from the 90s is due to underlying increase in computing power. I personally used a 50Mhz 486 in the 90s and now use a 8c/16t 5Ghz CPU. I highly doubt if we will see the same form of increase in the next 40 years

Or maybe not. Scaling AI will require an exponential increase in compute and processing power, and even the current LLM models take up a lot of resources. We are already at the limit of how small we can scale chips and Moore’s law is already dead.

So newer chips will not be exponentially better but will be more of incremental improvements, so unless the price of electricity comes down exponentially we might never see AGI at a price point that’s cheaper than hiring a human.

Most companies are already running AI models at a loss, scaling the models to be bigger(like GPT 4.5) only makes them more expensive to run.

The reason why internet, smartphones and computers have seen exponential growth from the 90s is due to underlying increase in computing power. I personally used a 50Mhz 486 in the 90s and now use a 8c/16t 5Ghz CPU. I highly doubt if we will see the same form of increase in the next 40 years

I doubt if this will make much difference. Offshoring as a tactic emerged in the pandemic when companies realised that being “remote” works just as well.

Sure, foreign R&D still gets amortized over 15 years (NPV ≈59 % of a full write-off, so you “lose” ~8.6 % of your R&D spend in present-value terms, and only 6.7 % of the cost is deductible in year 1, creating a 19.6 % cash-tax gap). But offshore wages are often 50–70 % below U.S. rates:

• Even after the slower amortization drag, hiring at half the cost nets you ~30 % total savings on R&D headcount.

• On a pure cash basis you only need ~20 % lower wages to break even; most offshore markets easily exceed that.

• So the labor-cost arbitrage far outweighs the tax timing penalty unless your foreign salaries are less than ~20 % below U.S. levels.

In short: the 15-year amort rule hurts your tax deduction, but 50 %+ lower offshore wages more than make up for it.

Sure, foreign R&D still gets amortized over 15 years (NPV ≈59 % of a full write-off, so you “lose” ~8.6 % of your R&D spend in present-value terms, and only 6.7 % of the cost is deductible in year 1, creating a 19.6 % cash-tax gap).

But offshore wages are often 50–70 % below U.S. rates:

• Even after the slower amortization drag, hiring at half the cost nets you ~30 % total savings on R&D headcount.

• On a pure cash basis you only need ~20 % lower wages to break even; most offshore markets easily exceed that.

• So the labor-cost arbitrage far outweighs the tax timing penalty unless your foreign salaries are less than ~20 % below U.S. levels.

In short: the 15-year amort rule hurts your tax deduction, but 50 %+ lower offshore wages more than make up for it.

You absolutely can. Just google for your “<printer name> compatible cartridges” , and you’ll find tons that are cheaper and work just fine with your printer.

This is going to have a catastrophic effect on closed source AI startup valuations. Because this means that anyone can copy any LLM. The person who trains the model, spends the most amount of money. Everyone else can create a replica at lower cost

OpenAI is going after a company that open sourced their model, by distilling from their non-open AI?

OpenAI talks a lot about the principles of being Open, while still keeping their models closed and not fostering the open source community or sharing their research. Now when a company distills their models using perfectly allowed methods on the public internet, OpenAI wants to shut them down too?

High time OpenAI changes their name to ClosedAI

I personally don't think that it is going to be a big issue. India banned TikTok, but it is still accessibly on the browser via a VPN. And everyone has a VPN installed on their phone these days. What is stopping people from just opening tiktok in their browser on their phone and continuing to browse their old accounts and content?

> Any modern card under $1000 is more than enough for graphics in virtually all games

I disagree. I run a 4070 Super, Ryzen 7700 with DDR5 and I still cant run Asseto Corsa Competizione in VR at 90fps. MSFS 2024 runs at 30 something fps at medium settings. VR gaming is a different beast

So when I can try this?

Not true for the entire C# ecosystem. I tried rebooting a Xamarin project I coded a couple of years ago. Had to spend weeks upgrading it cause Microsoft decided to discontinue Xamarin and force everyone to use NET MAUI

AI has boosted my productivity but only marginally. Earlier I used to copy paste stuff from stackoverflow and now AI generates that for me.

After years of being able to run SQLite on my mobile phone, my tv, my router and gaming consoles, I can finally run it on my browser. Which also happens to be running on the most powerful machine I own