Ask Paul: June 20 (Premium)

Happy Friday! Let’s kick off the weekend a bit early with another great round of reader questions.

? PC gaming, mainstream laptop

JustMe asks:

You’ve mentioned on Windows Weekly that some of the newer processor combinations make for decent gaming without the need for dedicated graphics. This could be a boon to travelers who need both a laptop for work but want the ability to game on the go. Is there any one CPU combination or a particular laptop series that sticks out as better than any other at this point?

Yes. Anything with a modern AMD processor–meaning “Zen 5” family–is better than anything from Intel, whether it’s Meteor Lake (Core Ultra), Lunar Lake (Core Ultra Series 2) or Arrow Lake (ditto). The integrated graphics in Lunar Lake are demonstrably better than those in Meteor Lake, so that’s good. (I can’t speak to Arrow Lake, but it uses an older integrated graphics chipset.) But anything from AMD is better than either. And there are tiers of AMD chips for premium laptops, so the newer and higher-end chips are the best of the best.

The asterisk here is that you can also get literal gaming laptops, portable workstations, or laptops aimed at creators that also included dedicated graphics. This will cost more, and the resulting laptops will be thicker, heavier, and louder, with more fans. But they will also perform better than PCs with dedicated graphics.

That’s not what I care about. I am blown away by the quality I see on AMD-based business and mainstream laptops these days. And if you end up with Lunar Lake, it works fine and looks fine. In a game like Call of Duty: Black Ops 6, I might see something like 1920 x 1200 at 30 FPS on low graphics quality settings with Lunar Lake but full resolution (2880 x 1800 or whatever) at 60 to 110 FPS on good/great graphics settings, depending on the chip and configuration. Both are playable. But the AMD versions are always a lot better.

? Only Apple could go to China and give them everything

JustMe asks:

Just finished reading Apple in China: The Capture of the World’s Greatest Company. My reaction was, “are you ?!#ing kidding me?!” Are you still planning on writing your review of this?

Yes. This book is a revelation. Brad and I discussed it a bit this past week on First Ring Daily. But I will publish a formal book review soon.

?‍? The AI paradox

wright_is asks:

I keep looking at the LLM chatbots and I still haven’t found a need for any on a regular basis. I use Perplexity for some searches and Brave’s Leo for others, but the actual ChatGPT and similar systems don’t bring any visible benefit for the work I do or my hobbies.

Curiously, I don’t use AI regularly either, and I often don’t even think to try. I can’t explain this per se. But I do know that my wife has only increased her usage of AI and that she does so to be more efficient and it’s working out well for her. She also recognizes that it’s only a matter of time before her clients start replacing her with AI. That’s a double-edge sword.

I won’t use AI to write. Writing is what I do. But it’s more than that, too. For me, at least, I will often start writing something and then it goes off in a direction I didn’t anticipate and leads to a new place. There’s probably a term for that, but it’s a form of thinking through something through writing, or whatever. I may think I have an idea or even a conclusion in mind. But then it takes on its own life.

Where could I use AI?

I use it for images on the site, of course. I use it small ways like grammar and spell checking. I have used it to upscale images. To edit images, I guess.

I could use it to summarize things, but rarely do. This is useful with videos, especially, and it could provide a hint about whether some part of a video is worth watching or whatever. I could use it to prioritize to-do items, perhaps. All kinds of things. I just don’t.

There is a business side to Thurrott.com that needs more attention. Is there some way to automate some of that using AI? That would be nice, a sort of digital assistant who (that) would just do things on my behalf to make my life easier. But that’s very vague. I don’t even know where to start.

Yeah. I can’t explain this. All around me, people I know–friends, family, normal, non-technical people–are using AI for all kinds of things. Some are even paying for it (and when they do, it’s always ChatGPT). But I rarely think to try to use it. No idea why.

On the other hand, we have companies like Apple, Adobe and a few others that eschew general LLM chatbots and instead put targeted AI features, often integrated into the UI, without making it explicit that AI is helping you complete the tasks – Apple has been doing this for years, as have Adobe, Luminar & Co.

This, to me, is how AI will really impact us. It will improve a million small things in a million small ways. Before Teams, I used to struggle with the monthly updates that Microsoft would add to Office 365/Microsoft 365 every month. When Teams arrived, that switched largely to new Teams features, and it was still overwhelming. And then AI happened, we got all these permutations of Copilot, and many of the new features we get each month are AI-based. It’s still overwhelming, but it’s also difficult to keep track of who gets what and where, which features are free, or maybe free but limited, and so on.

The issue here is that you won’t see this happening. Let’s say my wife text messages me. Basically any messaging app will have one-tap stock responses these days. Those are helpful. Are they “AI”? Or maybe you type something in that app and it suggests changes. If it’s a spell check-type thing, is that AI? What if it wants to re-write the whole text message? Is that when it’s AI? Does it matter? Or do these things just keep getting better, as they were before and would be anyway? Is this just marketing?

I don’t know. But I do know that many AI-based capabilities are useful across whatever apps. And that some are annoying. I’m not sure I’ve seen an AI summary at the top of Google Search that I’ve wanted, needed, or used, for example. But it could happen. Or maybe it could just answer the damn question. That would be useful.

Microsoft seems to be a bit all over the place, they are packing AI into everything, but it is often a chatbot style integration, not a “hidden” integration that just works and does something useful

Yeah. Microsoft has literally embraced its “copilot” concept as a side-by-side experience. But I always think back to when it added right-click support to the Office apps before adding it to Windows in Windows 95. And how amazingly useful that was: If you didn’t know which button to click or which menu to scan for a particular option, you could just right-click the thing you wanted to do something with and see the available options. It almost always offered the solution. AI should work like that.

Sometimes it does, I know. But I feel like the chat experience is perhaps best used as either a standalone experience or for those times when a more complex back-and-forth conversation is appropriate. If you’re going to reason out the solution to some problem, or the creation of whatever content, a chat interface can make sense.

I like the Apple, Luminar style better for the sort of stuff that I do. I don’t have to think about writing a prompt, I just select menu option or a style or click a button and the AI does its stuff in the background, just like every other option or button in the application. It is seamless.

Agreed.

But seamless isn’t “sexy”, if it isn’t in your face, a lot of people won’t realise that you have implemented the current hot technology in your products and they won’t talk about it, or even complain that you aren’t keeping up with the Joneses.

Yeah, if you don’t get that pink and purple animation and some kind of “hallelujah” sound effect, how would you even know that the AI has served your needs? Maybe it needs a tip jar. 🙂

There are certain areas where asking questions is relevant and can help in research tasks etc. But I think a lot of companies are getting lost in the “it must be a chat LLM to be sexy” hype, whereas seamless integration is more practical and more useful for many tasks.

The good news is that this will sort itself out. Remember, we’re talking about a company that added the Copilot icon to the Windows 11 Taskbar one month before shipping a major OS update, and then it went on to change the icon’s position three times. That was just an icon. Microsoft sorting through the “right” way to present AI-based functionality will take a lot longer, and the answer will vary by app/service.

At some point, it will have to do what customers want. Unfortunately for Microsoft, it appears that what customers want is some other AI, usually ChatGPT. One wonders about a future in which Microsoft might simply market its platforms as being so powerful by themselves (using its own in-house AI) that customers would never need to pay another company for those features. We wonder because they’re trying to charge us now, too, and these are the kinds of things we got “for free” as updates to the subscription we were already paying for.

E.g. dashboards, businesses have dashboards and the users want to see specific information every time. You could ask an chat LLM to get the same information for you, but the dashboard is constant and pulls the information in the same way every time, so you just need to glance at it, to see if there are problems. Having AI working in the background pulling in the information and putting it automatically in the dashboard is a better experience for most users, especially upper management who don’t tend to have the skills for using tech anyway, instead of them having to ask questions.

In fact, it could be a multi-layered experience, the dashboard with AI provides the status information and a prompt would allow the user to ask for more detailed information on specific items, allowing for traditional drill-down dashboards, but also the flexibility to ask for further information, once the drill-down has reached its limits.

Yes. I do like the idea of using AI to customize things like this. And this is where the natural language interfaces start to make sense. Describe what you want and get it. We’re all caught up in this “vibe coding” nonsense, but to reuse that terrible term, what you’re describing is almost “vibe managing” or whatever. This is a Star Trek-like future where you can simply ask a question and get the answer, with the range of both going from one word to infinity. And it can make things, like an interactive dashboard.

But in the future, there will be more complex, one-off, at the moment displays as well. Right now, you can open Google Maps and see which restaurants are in your area. You can fine-tune that view to show only those restaurants that are open right now. Are only 4.0 stars or more, or whatever. But in the future, this will be a temporary view, instantly compiled by your voice command, perhaps in a heads-up display you see via a pair of smart glasses or, God help us all, a corneal implant or whatever. Extrapolate that kind of thing out to all the little things we all do every day, and that’s the killer app for AI. Not one thing we all need. But the million things we collectively all do every day, the million tiny helpers that hopefully improve life.

? Microsoft, AI, and the future

Related to the above, will asks:

When Copilot launched a couple years ago, there was an excitement and anticipation around what this could do inside of the Microsoft ecosystem. Since then the product has changed multiple times and is spread all over the Microsoft stack in various flavors and features. However other AI systems have grown up and recently there is news that all is not happy between OpenAI and Microsoft. I am curious where you see Microsoft being with AI in a year or two from now? Even though they were early out of the gate, will they become eclipsed by others and become just something you use if you happen to, like Bing, but not the preferred choice for users?

There are all kinds of outcomes here. But I do think about this a lot.

I’ve introduced this idea of an instance in which Microsoft might lose to a competitor but still win. For example, if Sony defeats Xbox in the videogame market, but Sony uses Azure for its backend and/or game streaming service, and that generates enough income to offset that loss, is it still a loss?

In the AI space, Microsoft is building AI into everything it builds, and reusing the Copilot brand well past the point of common sense. But ChatGPT is by far the stronger brand and is far more popular. So popular, that it has become the Chrome of AI, if that makes sense, the product that everyone seems to use even though there is something built in to the platform that they get for free that does (basically) the same thing. But Microsoft takes 20 percent of all revenues generated by ChatGPT (and anything else OpenAI sells). If Copilot loses to ChatGPT, does Microsoft still lose?

My guess for Microsoft is that Copilot will remain as unpopular as things like Edge and Bing outside its core market, which is enterprises. And that it will be forced to offer more of the AI functionality that’s currently an additional paid (per use per month) Microsoft 365 Copilot subscription on top of the various Microsoft 365 subscriptions that these companies are already paying for. Because the alternative will be them using ChatGPT or whatever else. It’s going to have to make changes.

Microsoft 365 Copilot is too expensive, though it’s worth pointing out that ChatGPT Team is the same price. (And there are volume discounts to enterprises for both.) Quality is an issue, but to my mind, that gets ironed out, and the real issue is perception of quality/value. And that the ChatGPT brand, for better or worse, is so strong that even big businesses might adopt it instead of going with Microsoft, the single supplier. Renewed antitrust issues, with Teams in particular, do not help. Just bundling it makes sense. That may not be possible.

As noted above, the features that Microsoft is now adding through Copilot (whether it’s Microsoft 365 Copilot or just in general) were always the types of things we got “for free” as part of Microsoft 365 before (or Windows, in some cases). Charging extra for this is untenable for the business long term. The goal will shift to giving customers incentives to not go with a third party.

It may already be too late. Yes, many people will just use whatever AI is available in whatever app/service they’re using. Why wouldn’t you? Most don’t install a third-party spellchecker or whatever in Microsoft Word, right? But for more complex tasks, the things we’re describing now as “multi-agentic,” the brand and the perception of the brand matter. And that’s where ChatGPT and OpenAI become a real problem. They have that recognition. Like Tesla in electric cars before Elon Musk revealed himself to be not just an idiot but a dangerous idiot. Sam Alton is absolutely capable of a similar downfall, and OpenAI is almost certainly chaotic evil, not just chaotic. (Check out the OpenAI Files if you’ve not seen it.) But this shortcoming hasn’t hurt Meta yet. So all bets are off.

⌨️ Dude, you’re not getting a Dell!

wright_is asks:

Have you seen the new Dell Pro series laptops yet? Do you know how their build quality compares to HP and Lenovo ranges? Is Pro Lenovo L, Plus T and Max Carbon in Lenovo ThinkPad terms?

I haven’t, sorry. I’ve not reviewed a Dell laptop in many years. I saw a couple of them at Costco recently, and they seemed fine. But the designs are kind of generic for the most part, and it feels like Dell’s rebranding was specifically done to make them less interesting to consumers, which is a tiny market for them. But Dell is kind of unique in the top three PC makers in that its PC business delivers just half of its revenues, whereas PCs are by far the biggest business for both Lenovo and HP. I feel like getting out of the consumer PC market is the next step for Dell. And that it will retain the commercial side of that only because it’s a steady monthly revenue over a lot of time.

❤️‍? Gutsy

eeisner asks:

I always enjoy your takes on health and tech; thank you for your writing earlier this week.

I was particularly surprised by your doctor mentioning the kidney piece as the 3rd indication of health. I was assuming the answer would be gut health, which has recently become very trendy as of late.

Yeah. I was curious where she was going with that, and I feel like liver would have been the more obvious answer. But I guess that was her point: This third thing is a part of the puzzle few even think about, and it’s closely tied to the heart health (blood pressure/cholesterol) and glucose components in that each impacts the other and having a problem in just one will cascade problems elsewhere. I mentioned this to my wife, sister, and brother-in-law and all three dove into the health care apps on their phones to see (and then compare) their eGFRcr scores. Like me, none had ever heard of it.

Now, gut health can’t be measured through any health tech or directly through blood work (though a healthy gut leads to lower cholesterol), but I’m wondering how much this has been a focal point in your healthy lifestyle.

This is something I’ve wondered about, but the closest I got to this with my new doctor was in discussing fiber. I feel like I don’t get enough fiber, given that I stick largely to a low-carb diet. And I was curious if I needed to take some kind of a fiber supplement. I mentioned this to my wife, and she reminded me that I eat a salad for lunch five days a week (which she makes), and that is good. But that’s only when we’re in Pennsylvania. And it’s not like I’m going to start eating some high fiber breakfast cereal or whatever. I don’t really like most food items that are high in fiber, or don’t eat enough of them.

I will ask my wife about this.

? I abhor change

train_wreck asks:

Have you ever thought about making First Ring Daily into a weekly show? I kind of dropped off of watching them, as the content was getting a bit light. The most recent episodes this week were refreshing, and i think would make a good weekly show format.

I never intended for this to be a daily podcast, or something I do every morning, right at the time I should be writing. But what happened, happened. And here we are.

I don’t know. If this wasn’t Brad, I wouldn’t bother. It being daily is in the name, and changing that is not a good idea. And one thing I am acutely aware of is how much people hate change. For all those who may want this to be weekly, there are those who would miss it being daily. There’s no good answer.

I can’t believe I’m writing this, but I will soon be appearing in another podcast, which I believe will be weekly. More on that soon, but the episodes will be short in length (15–20 minutes) and usually with a very specific (tech-related) topic. With a co-host, most of you will recognize. Who came up with this idea. And I’m doing it for two reasons, even though I would have said before hearing the idea that there is no way I would ever do another podcast. I love this person. And it’s a good idea.

⚡ Thunderstruck

jrzoomer asks:

Paul, I’d like your opinion on Thunderbolt, the cable technology that was developed by Intel and Apple. I’m an avid user of a BenQ Thunderbolt Display, and I’m a huge fan of its technology and capabilities, but I’m puzzled as to why it hasn’t achieved broader market adoption. I love how my Thunderbolt Display simplifies cable management by consolidating connections into a single cable, acts as a versatile hub for peripherals, and enables seamless KVM-like switching between my Mac and Windows PC. Apple has included it in almost all of its Macs for a while now.

To me, this is more about Type-C as a port than it is about Thunderbolt specifically. And Type-C is something that works across Thunderbolt and USB, of course. This world is wonderful. But it’s also a mess.

In what is becoming a tradition in Ask Paul, I started writing an article called “Can We Please Fix USB?” a few weeks ago, the central premise being USB as a standard connection type and protocol has always been a good idea, but the implementation has always been inconsistent. And that the passage of time has only made that worse: The newer Type-C ports that are common today are used for both Thunderbolt and USB. It’s common to get a laptop with multiple Type-C (and Type-A) ports, each different in some way and with no clear or standard way to differentiate them.

There was a brief moment in time when the industry tried to solve this problem with Type-A ports by using colors. There were yellow ports, blue ports, green ports, and I think red ports. But not all PC makers were on board, and Apple certainly wasn’t, and that was that, and here we are. I would still like to solve this problem, which feels solvable. But there’s more.

There is also the cable problem. There is no way to know what USB or Thunderbolt capabilities that any given cable supports unless the manufacturer has clearly labeled it, which is rare. And so we all have drawers full of USB cables of various kinds, each different, with no way to know for sure which one you should use with which device. Including chargers, if that’s what you’re doing, and hubs and docks.

With regard to Thunderbolt specifically, I’ve been re-reading a lot of industry books over the past few months. There were several Microsoft books, and then most recently several about Apple and Steve Jobs specifically. And as I was re-reading the chapter about the iPod in his biography, I was reminded of this device, which I had, and of the early Mac OS X-based Macs in the early 2000s, and of Firewire (IEEE-1394). The original iPod had this gigantic Firewire port on its top to accommodate the high speed transfer of data from a Mac. When USB 2 arrived with faster speeds, Apple switched to that for future iPod models, easily availability on the PC.

Firewire was finally supplanted by Thunderbolt, a superset that included data transfer (PCIe), video out (DisplayPort), and power capabilities, plus chaining. The port, called miniDisplayPort, was smaller, and like Firewire, it appeared on Macs of that era. But that, too, was uncommon on PCs. And so Thunderbolt 3 and newer use the same Type-C port as USB. And they also work as USB, meaning that Thunderbolt 3+ ports can interact with basically any USB peripherals.

As the co-inventor of Thunderbolt with Intel, it’s not surprising that it championed Thunderbolt 3 and its Type-C connector in its products, and I figured this would lead to an industry-wide shift in which this just became the standard. But the PC industry had other ideas.

Beyond this, why hasn’t Thunderbolt become a dominant standard? And would you also agree with me on the benefits, or am I on my own island (with very few others) here?

The “why” part is simple: Cost. Thunderbolt is a set of specifications with a set of certifications, and while it’s been royalty-free since 2019, it still has an extensive certification process. And because it is royalty-free, the USB-IF was able to use it as the basis for USB4. So USB hasn’t just rolled over to Thunderbolt. In some ways, it’s a superset.

Because of all the cost-cutting, we have to deal with uncertainty. All device makers, but PC makers especially, cut corners wherever possible. Most phones still use USB 2 (5 Mbps) Type-C ports, though some newer and premium models (like the latest iPhone Pros) have USB 3 (10 Mbps) Type-C ports. Most PCs have some combination of Thunderbolt 4 and USB whatever. I have seen PCs with three or different USB ports, whatever combination of Type-A and Type-C ports, where each was literally a different type of port with different capabilities. Just looking at data transfer, you could have ports with 5 Gbps, 10 Gbps, 20 Gbps, and 40 Gbps transfer rates (and now 80 Gbps). And depending on the PC maker, they may be impossible to tell apart (and some may be Type-A).

USB4 closed the gap with Thunderbolt 4, in that they are basically identical. But of course we now have Thunderbolt 5, with double the transfer speeds (80 Gbps vs. 40). And so there is a USB4 version 2.0 (great name) with that speed as well. I’ve never seen either, but at these speeds, there are some important differences regarding cable lengths and cable types. But this is just more of the same: We have all these ports and cables that all look the same but have different capabilities.

Whether it’s Thunderbolt 4 or USB4 or both, the benefits to this system are obvious: You can have a single cable plugged into a computer that delivers power and, via a dock or hub, all the external peripheral connectivity anyone would want. A powered dock or hub used to be limited to 65 watts of power, which is just right for a mainstream laptop, but more recent versions can deliver much more than that (up to 240 watts, I think), and can power almost any laptop, even those with powerful dedicated graphics. This, to me, is the real benefit. We will always get faster/better/whatever capabilities in time, but the on cable thing is ideal. It’s the heart of what I think of as a More Mobile setup.

Related to this, is 40 or 80 Gbps even necessary for most people? No. And that, plus the lower prices, is why we see USB hubs of all kinds and USB docks, each with their own mix of capabilities. I have an Anker 555 USB-C hub (10 Mbps) that costs just $40 that does everything I need, and it real-world use, it’s just as good as the expensive HP Thunderbolt 4 dock ($330) that I also use. Technically, the latter is better, among other things, it delivers more power, but I never/rarely need that.

But the cheapness from device makers and all the resulting uncertainty is a black mark on this whole concept. And it’s why there is still some resistance to this stuff. This needs to be fixed.

Interestingly, Microsoft is trying to fix this. Good luck with that.