tpurves

This is such a dastardly psychological trick. Being slightly aswew really hard to fight the subconscious urge to reach out and 'fix' them. I almost want to rush out to a nearest McDonald's right now and buy one of these burgers so that I can make sure that it's buns are aligned properly....

Yep, I am in same boat. 32GB M1 Pro myself, and ya still a very solid machine. Thanks for that link.

So between AI and robots that can do this, humans are basically done right? I am going to bed.

Closest thing to a usecase I can think of are all military in nature. There's an argument that AI in space only begins to make any sense if the data source you are processing is also originating in space, and you need (for reasons) to run models on it in space before downlinking the results.

Maybe there's a very latency sensitive need to send realtime targeting information to a tomahawk missile in flight? But it's also too bandwidth, compute or cost intensive to send a firehose of raw spy-satellite data to a disposable one-way attack munition?

The data centers in space are actually for the spy satellites to use. That's all I got for practical applications.

It's not just chat. Remember image and video generation are on the table. There are already a huge category of adult video 'games' of this nature. I think they use combos of pre-rendered and dynamic content. But really not hard to imagine a near future that interactive and completely personalized AI porn in full 4kHDR or VR is constantly and near-instantly available. I have no idea the broader social implications of all that, but the tech itself feels inevitable and nearly here.

Either M5 Pro chips are finally here... or March is just the next time Apple can source any DRAM or SSD chips for the existing line!

In all seriousness, I am quite concerned for what the base model and upgrade pricing is going to be for the full M5 line, given the recent AI-driven DRAM supply apocalypse.

Ships of that era and leader had castles on both ends fore and aft. It's just the forward one than retained in usage as a sailing term, even after foredecks no longer looked like castles. The aft castle became a quarterdeck, a poop deck, a cockpit or a bridge etc.

Meanwhile, a built-up and elevated stern 'castle' is advantageous place to put the steering and command position, close to the rudder and with visibility of the whole ship, it's rig, plus where the ship is going. While maximizing mid-ship area for cargo. If you have to pick one end or the other, stern is the more comfortable end of the ship being most sheltered from wave action and weather. Being elevated and fortified also helps as a fighting/defensive position, but that is less important for modern cargo ships. 'Anticipation' isn't quite the right word as shipbuilders have always worked within the same basic design considerations and trade-offs, as the sea itself continues to enforce the same fundamental constraints.

[this is a reply to fourseventy] Looking up the violent crime rates by migrants in places like MN, it's effectively zero. As a rule, migrants and immigrants don't commit crimes at anything close to the rate of native US citizens.

Meanwhile in Minneapolis, the overwhelming majority of violent crimes (including aggravated assaults, theft, murders and sexual assaults) are being committed by ICE agents.

Your comment above should make more clear that the large scale healcare fraud you mentioned was not about Minnesota. It was nationwide, involved mass identity theft and large corporate scale white collar crime.

Dell is cooked this year for reasons entirely outside their control. DRAM and storage/drive shortages are causing costs of those to go to the moon. And Dell's 'inventory' light supply chain and narrow margins puts them in a perfect storm of trouble.

So Zen 6/7 will have a core design and a CCD design. But like past gens, these will be packaged into different products with different sockets and packages (everything from monolithic APUs to sprawling multi-chiplet Server cpus).

So to say that Zen 6/7 supports AM5 on desktop, doesn't necessarily exclude that Zen 6/7 product family in general doesn't support other new/interesting sockets on desktop (or mobile) also. Maybe products for AM6 and AM5 from the same zen family.

Medusa Halo and the Zen7 based 'Grimlock Halo' version might be the interesting ones to watch (if you like efficient Apple-stlyle big APUs with all the memory bandwidth)

Undoubtedly each new model from OpenAi has numerous training and orchestration improvements etc.

But how much of each product they release also just a factor of how much they are willing to spend on inference per query in order to stay competitive?

I always wonder how much is technical change vs turning a knob up and down on hardware and power consumption.

GTP5.0 for example seemed like a lot of changes more for OpenAI's internal benefit (terser responses, dynamic 'auto' mode to scale down thinking when not required etc.)

Wondering if GPT5.2 is also case of them in 'code red mode' just turning what they already have up to 11 as a fastest way to respond to fiercer competion.

Either way, you are right to point out that it important to only a try a pattern like this if your clients are highly trusted (or/and have additional compensating controls against DDOS threats). It would be beneficial if the OP made more explicit what their client/server relationships and also flagged the risk you mentioned for general audiences not to go implementing such a solution in the wrong places.

Thanks for calling this out. Here is a better comparison. Before Google was founded, the market for online search advertising was negligible. But the global market for all advertising media spend was on the order of 400B (NYT 1998). Today, Google's advertising revenue is around 260B / year or about 60% of the entire global advertising spend circa 1998.

If you think of openAI like a new google, as in a new category-defining primary channel for consumers to search and discover products. Well, 2% does seem pretty low.

What you are describing has been proposed before, for example within context of projects like Breakthrough Starshot. In that the case the idea is to launch thousands of probes, each weighing only a few grams or less, and accelerating them to an appreciable fraction of the speed of light using solar sails and (powerful) earth-based lasers. The probes could reach alpha centauri within 20-30 years. There seems to be some debate though about whether cross-links between probes to enable relaying signals is ever practical from a power and mass perspective vs a single very large receiver on earth.

It seems like strixhalo is a pipe-cleaner part of sorts and the real deal may have to be Medusa Halo. That one could be a monster. The bad news is that it sounds like it's a long way off (2027 sometime) so who knows what Apple M5 or M6 Max could look like by then for competition.

Around a decade ago, even as they were just launching Apple Pay, Apple was trading at a multiple barely over 10x. Street was valuing Apple like a manufacturing OEM company. I remember buying a small chunck of shares at the time thinking, this is crazy, just the services revenue off of owning these platforms is going to become massive one day.

America is trapped in a cycle where political parties have discovered that fear and anger drives voters whereas contentment with status quo does not motivate turnout. This leads to a scenario where parties will actively sabotage the resolution of painpoint issues such as immigration, healthcare, gun control etc. so long as it continues to create anger and fear that they can successfully blame on the other party. This behavior extends to voting against their own proposed policies in the interest of seizing/maintaining power over problem solving. And now deliberately creating crises (both real and fictional ones) has become the game-theory dominant strategy in American politics.

You forget, they also clean-sheet designed the starliner starting in 2014 and that project... also happens to exactly prove your point. (at least 2B over budget, and 8 years after it's original operational target of 2017, has yet to fly a fully successful mission)

Did you take a job at Apple just so you can accept one or more of those PR's that's been bugging you? :)

Wasn't there an xkcd about that scenario...

And 75% of that just gets shipped right over to nVidia as pure profit. The mind boggles at the macro-economic inefficiency of that situation.

It's really both effects happening at once. AI is just like the invention of the assembly line, or the explosion of mass produced consumer packaged goods starting from the first cotton gin. Automation allows a massive increase in quantity of goods, and even when quantity comes with tradeoffs to quality vs artisanally produced goods, they still come to dominate. Processed cheese or instant coffee is pretty objectively worse that the 'real' thing, but that didn't stop cheap mass production still made those products compelling for many million/billion of consumers.

You can still find a tailor who will hand make you a bespoke clothing or sew your own clothes yourself (as even the boomer generation often did growing up), but tailored clothing is a tiny fraction of the amount of clothing in circulation. Do tailors and artisanal cheese makers still exist? Yep, they are not extinct. But they are hugely marginalized compared to machine-made alternatives.

Card networks have the most reliable protections and guarantees, their systems and rules are designed around consumer trust in their brand and zero-liability (to consumers).

Yes. When he's right, he's right. However the main issue I have with GN is how Steve tends to go full Leeroy Jenkins pitchforks and torches for 9 out of every 5 actual scandals in the tech industry.

Arbitraging fraud has always been PayPal's game. For the mass market of small-scale sellers there's a threshold, whre the cost of onboarding, verifying, underwriting and eating fraud exposure is greater than the revenues you could ever expect to make from them at their low volume. So these segments went un-served, same with many other segments of transactions or transactors. PayPal's model has always been to accept a higher risk tolerance, then to have just-enough compensating controls and/or liability-dodging to make that work. At least just well enough to have net-revenues bps exceed their fraud-loss bps.

And when that balance is working, it kinda works, but when it doesn't...

And incidentally, it's not just PayPal with the fraud problems these days. It's everybody in the banking and payments space. AI is so far quite asymmetrically helping the bad guys more. It's bad out there.

No you were right to be disappointed! I was similarly in a state of excited awe to see someone play actual quake inside the notch of a mac. Much like the noble hacker tradition of getting Doom to run on the most unexpected devices possible. So yes, ID Software games running in weird places is an entirely reasonable expectation!

'just a terminal' is maybe also a legit term of art and entirely useful for somebody out there. But much less cool that we were hoping for!

It's a distraction to try and imagine homomorphic encryption for generic computing or internet needs. At least not for many more generations of moore's law and then even still.

However, where FHE will shine already is in specific high-value, high consequence and high confidentiality applies, but relatively low complexity computational calculations. Smart contracts, banking, potentially medical have lots of these usecases. And the curve of Moore's law + software optimizations are now starting to finally bend into the zone of practicality for some of these.

See what Zama https://www.zama.ai/ is doing, both on the hardware as well as the devtools for FHE.

Graduate employment crises happen at least once every decade (initial covid, 2008 crisis, dotcom bust, early 90s recession etc). What I could argue is more unusual are those bubble times when getting a great job right out of college is easy. And then how quickly those expectations are taken for granted...

GenX'ers will remember the days of 'Slackers' 'Reality Bites' and the malaise of those who graduated with fancy degrees in the early 90's but stuck in barista jobs etc.

That implies the interesting question though, which shape and mass distribution comes closest to, or would maximize relative uniformity?

Well, lifeforms on earth are all pretty dependent on being water based, and water in the liquid state specifically. Maybe there is a possibility of exotic life based on some other types of chemistry and/or phases of matter. But the fact that earth happened to form in this particular goldilocks zone for water-based life is probably why that's the only life we can see for now.