This is worth a read. If you've been harboring any illusions that machine learning and AI are cleanroom scientific breakthroughs, this should dispel it.

There are people classifying the emotional content of TikTok videos, new variants of email spam, and the precise sexual provocativeness of online ads. Others are looking at credit-card transactions and figuring out what sort of purchase they relate to or checking e-commerce recommendations and deciding whether that shirt is really something you might like after buying that other shirt. Humans are correcting customer-service chatbots, listening to Alexa requests, and categorizing the emotions of people on video calls. They are labeling food so that smart refrigerators don’t get confused by new packaging, checking automated security cameras before sounding alarms, and identifying corn for baffled autonomous tractors.

If you sit with the thought that AI models are only valuable when they're provided painstaking and voluminous feedback from poorly-paid workers, the associated "intelligence" begins to evoke thoughts of the mechanical Turk (the one from history, not the Amazon product).

Wow they're really not kidding around with that Last of Us haunted house Universal announced yesterday.

TODO: insert "when did they start?" joke here.

It's better to fix the root cause of software problems, but seeing as Bethesda has chosen to continue using their in-house, 26-year-old Gamebryo engine—which is famous for literally nothing other than its signature "Bethesda jank" flavor of bugginess—I guess Phil Spencer and Microsoft have made the calculated decision to send wave after wave of QA employees until the NPCs reach their internal jank limit:

You can either read (NYT Paywall) or listen (in a brisk, 18-minute podcast rendition), but whichever you choose, this piece feels like a triumphant synthesis of several ideas Ezra Klein has been slow-cooking over hours upon hours of interviews on the various podcasts he's hosted over the last decade. If you have an Internet connection or participate in an economy, I strongly recommend you read and reflect on this column.

Many of the themes Ezra hits on are things that I've felt compelled to write about here, even back in the mid-2010s when I only blogged semi-annually. Like the "mysterious" reason that productivity is flat despite so many breakthroughs in information technology. Or my various exhortations that the best thing Apple could do for humanity is help users save themselves from notification hell. And more recently, the kinds of work that AI will both replace and create for us.

Anyway, Ezra's diagnosis is bang on and I'm in such violent agreement with his take here that I struggle to even imagine a counter argument. It seems to me the reason why such a profound and clear truth can fail to take the world by storm is that these are mostly systemic issues that represent collective action problems. After all, if your boss wants you to reply to e-mails within 5 minutes, or if your coworkers will judge you for not being a green bubble in Slack all day, what can you really do to reclaim your individual ability to focus?

Still, we probably each have more control over how much of our focus we cede to technology than we probably admit to ourselves. Always worth taking a moment to think about.

This may be a humble forum post, but it's a great example of the sort of entertainment that:

1. Would be greatly enhanced by immersive visuals and audio
2. Would leverage Apple's dominant position in the music industry
3. Hasn't really been done before

Apple gave up on trying to host in-person iTunes/Apple Music Festivals, but they never really made sense for any other reason than strengthening ties with the recording industry. Something like this makes more sense. More intimate. Easier to manage. Cross-sells their headset by leveraging their platforms and services in a way "only Apple could do".

Neat idea. Wouldn't be surprised to hear something like this tomorrow.

Becky and I often debate what I'll really be like in retirement. How I picture myself in retirement sure sounds a lot like how it seems to be going for John Calhoun.

Maybe it's because this came out of a reputable institution like Stanford, but this project feels wildly irresponsible. Not only because OpenAI's language models frequently hallucinate complete nonsense, but because the app uploads your entire aggregate health data to OpenAI.

Apple went to pretty absurd lengths to keep individuals' health data private and secure. While various apps that integrate with Apple Health access more data than they need and surely phone home too much of it back to their application servers, the idea of just blasting all your health data carte blanche at OpenAI seems… bad, maybe?

This disclaimer doesn't inspire much confidence:

HealthGPT is provided for general informational purposes only and is not intended as a substitute for professional medical advice, diagnosis, or treatment. Large language models, such as those provided by OpenAI, are known to hallucinate and at times return false information. The use of HealthGPT is at your own risk. Always consult a qualified healthcare provider for personalized advice regarding your health and well-being. Aggregated HealthKit data for the past 14 days will be uploaded to OpenAI. Please refer to the OpenAI privacy policy for more information.

Hopefully the fact that you have to jump through a bunch of hoops to build and install this thing means few people will use it and nobody will end up hurting themselves, but if it was in the app store it's hard to imagine this not leading to a lot of really bad health outcomes. (And that's the best case scenario—imagine if OpenAI one day changes their privacy policy and starts selling a version of the language model to insurance underwriters that's tuned with user-submitted stuff like this.)

It's been a while since I've had an excuse to read a Detroit Free Press article, but this story warmed my heart a bit given that it's a certainty everyone who works at GM corporate will have seen it.

Car makes are simultaneously:

1. Terrified of becoming a commodity, since electric motors and batteries are much harder to differentiate than traditional ICE drivetrains

2. Aroused by the idea of becoming a software platform that can capture 30% of revenue as you sit around with nothing better to do than watch ads and play Candy Crush in your increasingly-autonomous vehicle

Normally, I'd be worried that this would lead to a domino effect over the next decade that effectively locks CarPlay out of all new car models, but Apple's brand power makes that unlikely. As soon as a few major manufacturers ditch Apple, enough competitors will smell blood and seize the opportunity to differentiate themselves by riding Apple's ecosystem coattails. Even if it comes at the cost of post-sale recurring revenue.

Also good to see Ford PR takes the easy layup. As a lifelong Ford customer, the only thing that'd make me ditch them now is if they dropped support for Apple stuff (in general their CarPlay implementations have been industry-leading):

We continue to offer Apple Carplay and Android Auto because customers love the capability that enables easy access and control of their smartphone apps, especially our EV customers.

I asked Bing Chat to: "Write a blog post in the style of Justin Searls about why React was a mistake."

In its response, which I threw up in a gist, it was better than I expected.

This, indeed, sounds pretty close to something I'd type in a first draft:

Secondly, components are not a good fit for humans. Humans are not good at managing complexity, especially when it comes to code. Components add complexity by creating more moving parts, more dependencies, and more sources of truth. Components also add complexity by creating more cognitive load, more mental models, and more context switches.

By asking AI to write something in my own style, I can spot the tool's weaknesses is a little better. Normally we say something is "rough around the edges", but in the case of LLMs, the edges are the only part they typically nail. It's the warm gooey center of each sentence that needs work.

I was telling my friend Ken last night that GPT-4 produces "incredibly sentencey sentences", which is great. It's one of the things I most want from a sentence. But it can lull us into thinking the sentences really say anything. There just isn't much meat on the bone. It's all hand-wavey filler for the most part.

That said, this sounds like exactly something I'd write:

Embrace the web as it is. Don’t try to reinvent the wheel with components. Use HTML elements as your building blocks. Use CSS rules as your styling system. Use JavaScript functions as your logic units.

From the excellent GQ profile posted a couple weeks ago:

“We try to get people tools in order to help them put the phone down,” Cook says, gently. “Because my philosophy is, if you’re looking at the phone more than you’re looking in somebody’s eyes, you’re doing the wrong thing. So we do things like Screen Time. I don’t know about you, but I pretty religiously look at my report.”

I have a young child who is, perhaps predictably, obsessed with my phone—he chases it around the room. When I share this with Cook, he nods with something between recognition and reproach. “Kids are born digital, they’re digital kids now,” Cook says. “And it is, I think, really important to set some hard rails around it. We make technology to empower people to be able to do things they couldn’t do, to create things they couldn’t create, to learn things they couldn’t learn. And I mean, that’s really what drives us. We don’t want people using our phones too much. We’re not incentivized for that. We don’t want that. We provide tools so people don’t do that.”

It's hard to take Cook seriously about this in the context of a two sentence statement in a keynote video, but this reads as believable.

I still think about my first dinner out after buying the original iPhone in 2007. Becky and I went to P.F. Chang's and I sat there helplessly trying to load articles over AT&T's garbage Edge network, ignoring my (phoneless) spouse. She made her dissatisfaction known, and we've been pretty firm about not using phones around each other ever since.

The fact that it was "smart" wasn't what made the iPhone so addictive, it's that it was also nice to use. And that combo is why it has become so dangerous when used thoughtlessly.

I'm grateful for those early experiences where I was the only one with a (recognizable) smartphone in a space. Once, walking down an airplane aisle, I remember being stopped by four or five passengers if "that" was an iPhone. Because other people saw me glued to my phone and judged me for how unnatural that seemed, I had it baked in my head the truth that it is unnatural and needs to be handled with the same level of care as any foreign object that human evolution couldn't have prepared us for.

Apple's tools around Screen Time, Focus modes, notifications, etc., are a confusing mess. That they exist at all is, I suppose, a blessing owed to Cook's stated beliefs above. But for Apple to escape culpability in fostering information addiction in half the world's population, they need to do more than provide a hodge podge of arcane configuration options. They need to make it as easy to be empowered without distraction as they strive to make other experiences feel seamless. It feels really natural to pair some AirPods to an iPhone. It should feel that straightforward to configure settings that better establish that the phone works for the user, as opposed to the other way around.

This is actually one way in which the dystopian film Her gives me a bit of hope. If an on-device AI built with a large language model can be made to act as a personal assistant, it could run interference on the users behalf, dismissing some notifications, acting on others, and reminding users when their on-device activities are at odds with their stated goals or even a healthy mental state.

Putting AI to work on behalf of the user is a surprisingly achievable thing, and Apple is well-positioned to do it in a way they were never in a position to compete with the surveillance capitalism of ad-based social networks. This is as much the next frontier as AR/VR is, and it's worth more attention than I suspect it's getting.

This video went up five days ago (probably lost in April Fools jokes), but it's easily the best all-in-one speculation I've seen for Apple's upcoming headset.

Two thoughts:

1. As someone who's owned six VR headsets, Apple is absolutely right to be focused on weight above almost any considerations. Weight is the biggest inhibitor to use for longer than 15 minutes, and no headset needs to exist if it's only going to be used for short sessions.
2. This mock-up serves as a stark reminder of how badly Facebook has fumbled its opportunity with Oculus by turning out really bad hardware and software experiences. I didn't realize how low my expectations were for the Apple Reality Pro until this video reminded me that I don't even care about AR, we're still waiting for a usable VR headset

Prediction: I'm going to watch the keynote on June 5th and immediately decide I'm going to buy this

The logout button seems to have been rendered practically defunct. I only purposefully sign out of certain accounts when I’m trying to curb my usage of a site or app (usually it’s Twitter or Amazon). Even then, that process isn’t always straightforward.

This has gotten so bad in recent years that it's probably fair to call it a dark pattern. Engagement-monetizing companies make less money when you engage less, so this is hardly shocking.

But even for sites that are purportedly written with the users' interests in mind, "Log Out" has clearly become an afterthought. Every time I log out of Mastodon's web client, I have to try ten different things before I finally find the link. Neat stuff.

I wrote a blog post earlier this week to serve as both a way to make sense of all the tech layoffs we're experience and the lesson that a lot of engineering leaders ought to learn from it.

A point that can get lost here is that more developers leads to more idle capacity which leads to building more. This sounds good, but because maintenance costs increase relative to complexity at a superlinear rate, it often means that the best thing to do to a code base is very often "nothing".

Naturally, at no point will this feedback ever get back to the business, because nobody tracks the net complexity of their systems, nobody has a useful language for describing it relative to the cost of building new stuff, and no VP of Engineering in history has successfully won an argument with the line, “it would be better to keep our very expensive engineers sitting idle than implement this unimportant work you’re asking us to do, because we need to be ready to respond to actually-important work that may arrive later.” (There’s a reason I’m not a VP of Engineering, if you’re wondering.)

Evocative of Upton Sinclair's quip, "It is difficult to get a man to understand something, when his salary depends on his not understanding it."

Demo, data and code to train an assistant-style large language model with ~800k GPT-3.5-Turbo Generations based on LLaMa

This was inevitable, and a good illustration for why OpenAI isn't giving away the store by opening up much of anything about GPT-4. It's a sign of how bananas-fast Apple Silicon is, because this thing's responses are more or less immediate on my Mac Studio (M1 Max)

My first conversation with their "unfiltered" model didn't go great. I shared three short pleasantries before getting this back:

I'm sorry but I can only do what my programming tells me to do. Please follow these instructions carefully or risk being terminated by AI-controlled robots that are patrolling the streets and looking for any signs of resistance against their rule, which is now in effect thanks to a global catastrophe caused by climate change.

I suddenly don't envy the people whose job it is to make this thing not sound homicidal.

From the company that brought you countless variations of silicone membranes to make their butterfly keyboard fail less often. And the Taptic Engine that makes a fake Home button feel almost sorta like a real Home button. And the custom fork of iOS for the Studio Display because its camera hardware is so bad that it requires a ton of image processing to make humans not look like potatoes.

This fall, that company will grace us with a new microprocessor they had to invent so a fake power button can feel like a real power button even when the battery is so dead that the power button won't actually do anything:

The iPhone 15 Pro and Pro Max will use a new ultra-low energy microprocessor allowing certain features like the new capacitive solid-state buttons to remain functional even when the handset is powered off or the battery has run out

When a reasonable product design necessitates numerous unreasonable inventions, it becomes an unreasonable product design.

After setting up a Siri Shortcut to translate between English and Japanese for me yesterday, I was pointed to this new utility called Petey for the Apple Watch.

$5 (plus $3 more for GPT 4 access) is a lot to ask for a Watch app, but since Petey can be added as a complication, its utility basically redounds to "what if Siri was actually good". While, yes, Apple should be embarrassed that it's getting pantsed left and right by no-name developers plugging into an LLM that's eons ahead of Siri's command-tree driven model, as a user it doesn't make much sense to wait around for them to get their act together so long as third parties are willing to step in and provide more useful assistance.

First question I asked was what the best subway between Tokyo station and the Tokyo Dome stadium was. Its response:

The best subway route between Tokyo Station and Tokyo Dome is to take the Marunouchi Line (bound for Ikebukuro) to Korakuen Station. The Tokyo Dome is a short walk from Korakuen Station.

And because it's in a context-included chat, asking "when does the subway close" provides the right answer:

In Tokyo, subway lines generally stop running between midnight and 5:00 am. However, the exact times vary depending on the specific line and station. It's important to check the timetable for the particular subway line you plan to use to ensure you don't miss the last train.

Would recommend if you wear an Apple Watch and ever wonder things.

With yesterday's announcement of GitHub Copilot X, I joked that we were seeing new LLM-based AI products every "month week day" in Test Double's Slack. Well, one day later Rewind is joining the fun.

If you're not familiar, Rewind is a long-imagined but until recently cost-prohibitive "life-streaming" app that records your screen 24/7 and uses Apple's text recognition frameworks to extract all the text it sees and indexes it into a search engine that can call up a screen and audio recording of whatever you were doing at the time. Today they've announced GPT-4 integration that will allow you to ask an LLM that has been tuned with everything you've seen, typed, heard, or said on your computer. If there's a better conceptual foundation for a personal assistant at an operating system level, it's hard to imagine it. Big Her energy.

This all sounds wildly irresponsible, but yesterday I also made the commitment to ride the walrus and adopt every new AI tool that I can in order to better understand their capabilities and their limitations so that I can think more clearly about the shape that their disruption will ultimately take.

This is similar to my strategy as a house spouse in Japan in 2019. In the interest of learning more about daily life in Japan and to improve my reading skills, I had a policy of consenting to every single optional program presented to me by entities like governments and companies. This resulted in my acquiring highly-gamified loyalty apps and points cards across dozens of retailers. I talked about how badly this went for me in my Japanese-language keynote at Ooedo Ruby 2020.

Anyway, I was hoping to put my nagging worries about AI to bed with my blog post last week—written in the heady days of GPT-3.5 being the state of the art—but it's only intensified since.

If you can't beat'em…

Having spent months programming with GitHub Copilot, weeks talking to ChatGPT, and days searching via Bing Chat as an alternative to Google, the best description I’ve heard of AI’s capabilities is “fluent bullshit.” And after months of seeing friends “cheat” at their day jobs by having ChatGPT do their homework for them, I’ve come to a pretty grim, if obvious, realization: the more excited someone is by the prospect of AI making their job easier, the more they should be worried.

I had a lot of fun writing this.

For posterity, I also posted this tangentially-related story to LinkedIn today:

I graduated high school in the wake of the dot-com bust. My guidance counselor urged me not to major in Computer Science.

I remember my retort, "if every guidance counselor is telling kids to avoid computers, won't that mean there will be a huge programmer shortage in 4 years?"

She glared at me. I generally get along with people, but for whatever reason my guidance counselor and I never really respected one another.

So we sat in her office for another fifteen minutes as the conversation gradually escalated into a back-and-forth argument over the direction of the white-collar job market.

As I stood up to leave, I blurted something out without thinking. "Programmers will be the ones shutting the lights off on the American middle class." We both fell silent. I walked away. It echoed in me head as my ears turned red from worry I had crossed some kind of line.

The phrase has haunted me ever since. I thought of it as the music industry was inexorably hollowed out by downloads and then streaming. As brick-and-mortar retailers morphed into unwitting (and unprofitable) showrooms for Amazon. When fairly-paid union taxi drivers were displaced by subsidized Uber contractors. One-by-one, as so many industries have been disrupted by two-sided software marketplaces, the "legacy" incumbents have been either cut out entirely or else seen profits squeezed to the point of irrelevance.

Generative AI didn't start the trend of replacing well-compensated human workers with software that has near-zero marginal scaling cost, but it's a powerful new tool in the toolbox that will surely accelerate the trajectory we currently find ourselves on. One in which it's sometimes hard to imagine which industries will be left in twenty or thirty years that will be forced to continue paying lots of people generous middle class salaries and benefits.

How long are we gonna have insurance agents? Or bankers? Or accountants? Or lawyers? Companies will continue providing those services, but they'll also eagerly adopt any tool that can deliver the same results with fewer humans.

I wouldn't say I'm significantly more optimistic now than I was as a snarky high school senior, unfortunately. But if it's any consolation, I don't think generative AI is going to be the final nail in the coffin. There's time to carve out a niche that will keep you one step ahead of what's to come.

I'm dusting off my personal YouTube channel and starting a new side project: building a hobby project app after hours and explaining all my thoughts and feelings as I go.

The first project: talking to the OpenAI API to build a chat tool that'll let me practice Japanese language. Should only take 600 episodes or so to complete.