robot_jesus

That is such a refreshingly fair and honest assessment. Thanks for sharing! To me, that engenders a lot of trust in what you’ve built.

> For example, I worry that biology will have a strong attacker-defender asymmetry, where sufficiently capable models may be able to quickly weaponize pandemic-level viruses with widely available materials, whereas defense against these agents is a multi-year operational task in the best case (as we saw with Operation Warp Speed)5. Questions like this should be empirically answered by rigorous pre-release testing, not assumed in advance.

One of two outcomes are true in this scenario: 1) this is unrealistic fear-mongering or 2) we're all fucked.

I just don't see how our solution to this can be export controls or bans. If the fear of a bioweapon engineered by an open weight AI is a legitimate threat, our only option is to develop effective countermeasures (and perhaps the same AI will assist with protecting). Open weights are not going away and pretending like some sort of "ban" will prevent bad actors from accessing them is a fairy tale.

Some real "You wouldn't download a car" vibes at points in this thread.

Is there a way to assign certain domains to always open in a certain container? e.g., if I type the url of my bank that it immediately becomes a banking container? That would be great.

I’m intrigued by the power of these models to help me take more control of my healthcare.

Sure, doctors’ advice is best, but sometimes I wait 6 months to get to see a specialist who seems hell bent on getting through our appt as quickly as possible.

Models are infinitely patient and can see a lot more context about my health.

But, the privacy is a big question mark. The page extensively reassures on privacy (no model training, encrypted, no ads, can be deleted). There’s only one part that stands out to me:

> Memory remains in your control. To personalize future chats, memories can be created from your Health conversations. Memories are not created directly from connected medical records and Apple Health information. If you’d like to have a conversation without creating memories, you can use Temporary Chat (opens in a new window) or turn memory off in Settings > Memory.

Great that you can turn it off in settings but I wonder how many people will do that and how much sensitive context leakage will come out of your conversations and suddenly you’ll get ads for rashes and whatnot.

There's a lot to unpack here, but a few thoughts. SpaceX wasn't added to the Nasdaq until two days ago, which is right around the time the stock was reaching it's current lows. Doesn't make sense to say that they prices exploded because of that.

Second, I don't know the details but it's extremely unlikely that most SpaceX folks have been able to sell any shares. Typically there's a 90–180 restricted window where pre-IPO holders are not allowed to sell.

The limited float could theoretically drive up, but as I noted, it wasn't added to the Nasdaq 100 until two days ago, so it's not something we can really attribute to that.

Btw, I loathe Elon Musk as a human, but am just clarifying some facts.

Weird, right? Especially on a website with an owner who so ardently supports free speech. </heavy sarcasm>

It's a fair callout, but for me (and increased propensity for Alzheimer's) the calculus is a bit different. Making to to my mid-70s with full mental faculties would easily be worth the tradeoff of an increasing likelihood of shingles at that point. And besides, I would presume I would be able take the vaccine again?

I'm in my 40s with genetic predisposition for Alzheimer's. Been seriously considering the past year or two paying out of pocket for Shingrix. I think it would be ~$500 total for two doses.

Sure, I could wait 7 or 8 years until I qualify via insurance, but is that really worth the risk for what is an easily absorbed cost to me? Especially when I have a friend in her late 30s who just went through a very rough bout of shingles?

It makes sense to have targets like age 50 for population-wide public health recommendations. But it can and does infect people of much earlier ages.

Recent articles like this make me think I'll go ahead.

Same. I really am curious if either 1) I am using it in genuinely different ways or 2) these people are being willfully disingenuous.

Neat project and I love the motivations behind it.

A few friendly recommendations:

• The images on the results pages are in desperate need of optimization. Plenty of 500KB+ thumbnails and even a couple 2MB+ thumbnails I saw in there. Could easily be optimized 90%+ with no loss of quality

• The instant live search can be a little distracting, particularly paired with the heavy loading of the thumbnails

• I know you don't want to create your own full ecomm site, but even just a hover PDP without having to click off could be nice, if you could pull in the key product details. I know the goal is to support the destination sites but it was a lot of back and forth to me

• Any ability to validate that a product is available vs sold out (and note that on the results pages) would be appreciated. Probably 75% of the items I checked on ttgaming.quest were sold out. A banner across items not in stock would be helpful. Or a filter on the search results page.

Keep up the good work!

Personal prediction: I do think the market will essentially force their hand to include it in subscriptions before too long. OpenAI, local models, Chinese models will continue to improve.

But, there are also harsh realities of compute volume and cost to run all of these will be fighting against.

What I do expect is a multi-tiered rollout of future models. You want the latest SOTA release? Usage credits.

Subscription plans will end up getting models on a lagging interval of a few months.

I love stuff like this. Even though I don’t have a machine capable with running OS 9 natively, I’m glad this exists. Looks awesome!

Thanks, yeah that's past my personal limit for a very niche project. I like the concept, though.

I was looking around but either I missed it or it’s not spelled out. Do you recall a ballpark cost for the components? I didn't feel like individually pricing out the many components.

You and me both, but I also recognize others disagree so ultimately, we'll see what the market decides.

Apple's annual gross profit was $195B last year against an R&D budget of around $35B. So, they've got more than enough spare change to throw around. I'm sure whatever they're spending on foldables isn't impairing them financially in any way.

I'm more concerned for what it means for focus, fragmented ecosystems, user experience, etc.

From Jobs: "People think focus means saying yes to the thing you've got to focus on. But that's not what it means at all. It means saying no to the hundred other good ideas that there are. You have to pick carefully. I'm actually as proud of many of the things we haven't done as the things we have done."

They have some for sure for iCloud. Do they have enough to handle this volume of compute AND is Gemini allowed to be run on those? That was more what I was questioning/curious about.

Yeah. It's clear they've been hearing the complaints. Not just Liquid Glass, but they even talked about the inconsistent menu bar icons and problems with rounded corner radii (among a bunch of improvements). I'm excited that this is basically Snow Leopard part II, for those who remember.

I agree it's a funny look, but my guess is that it comes down to the cross-border data transfers and non-EEA tech providers. So even if Apple has private cloud compute and is using Gemini models, there are probably a lot of legal hoops to jump through and/or European-based data centers to spin up?

Parent comment urged everyone to cancel NYT subscriptions and the child comment respectfully disagreed and explained why they still find value. That seems like a very direct response. These aren’t formal arguments and rebuttals; just opinions. I appreciate hearing both perspectives.

Agreed 100%. But for me it was many, many years ago. But this is my daily shoelace knot and it works amazingly well.

Pointless article (like much of the AI marketing hotness and spin room).

> The new valuation is nearly three times higher than the company’s February valuation, when Anthropic was estimated to be worth around $380 billion.

> In March, OpenAI was valued at $852 billion following a record $122 billion funding round.

Basically, today (Late May) we're declaring Anthropic the most valuable. They've nearly tripled in value since February. But also, OpenAI was $852B in March and presumably has grown since then.

In a few weeks we'll either have a new rounding of funding for OpenAI or they'll announce their IPO and the hype train will be abuzz that they're now the most valuable.

Or, "Person who sells tokens responds to article claiming MCP spends too many tokens says please keep buying tokens."

An incendiary clickbait title, and indulgent writing style turned me off. I understand the author’s point and to an extent agree. But I also can’t help thinking that these people plant a flag in the sand while the world just moves on by, leaving them increasingly out of touch.

>”This is why you can still ask an AI to tell you about the scene in VS Naipaul’s Dashed Against the Rocks in which a donkey is thrown from a hot air balloon”

Well, I did ask Gemini (3.1 Pro) that question verbatim and it wasn’t fooled in the least. People who rail against LLMs for hallucinations like that always read to me like people who haven’t used it since writing off chat got 2.5.

Braess's paradox for roads. When we add capacity to road networks, traffic increases even more than the capacity.

https://en.wikipedia.org/wiki/Braess%27s_paradox

I love my em dashes but I can't disagree with your assessment at all.

The entire tone is pure AI slop.

I struggled with the battle system, as well. Block was way too difficult with a mouse on desktop.

Keyboard buttons would be better or slightly more windup to the attacks to allow more reaction time.

But, I still found some tactics. Similar to OP, I'd send two sacrificial lambs up and focus entirely on a single decently-high (say, 70+) leveled-up number (minimum level 3). Let the other two numbers die, block on the focused number, heal as needed and hit that Divide as soon as possible. With 6 or 7 hearts I was able to finish levels 10 and 11 that way.

Confirming the same error on my side.

This is the only (very small) blemish on an otherwise fantastic game.

I know this game is a satirical sendup of gacha, but in the same way that Universal Paperclips subverted the clicker game genre and made something fantastic, I find this stripped-down gacha utterly charming. Thank you to the dev!

The "Four lessons" that came out of running this work at scale made me chuckle. Three of the four were essentially identical and entirely obvious. In short: specific, narrow requests work better than "find vulnerabilities." Well, d'uh.

But, I did think the adversarial review (while not novel at all and talked about much in HN circles) is interesting and distinct, at least. I need to put this to work in more of workflows. I think it could be beneficial for non-coding tasks, too.

https://blog.cloudflare.com/cyber-frontier-models/#what-a-ha...