sangsara.net

Tag: Generative AI

  • Week 21.23 (poem edit)

    An AI turned this week’s notes into poetry.

    A Chronicle of Week Twenty-One

    In a week where work did reign,
    Much to tell there’s little gain,
    Round it though, we gently dance,
    For work’s secrets shan’t have chance.

    (more…)

  • Week 21.23

    It was one of those weeks where not an awful lot happened outside of work. I don’t talk about work here but let’s sort of circle it.

    Reflections on AI

    One thing I can say is that I started making a presentation deck about the use of generative AI and GPT in design, initially to share my findings so far but increasingly as an exercise in structuring my thoughts into anything at all useful.

    A couple of things, on reflection: an AI assistant or collaborator poses significant risks for a lazy human in certain tasks since it tempts us to quickly accept its output without evaluating potential improvements. Assuming AI can do a job to 90% of a human’s rigor and quality, figuring out what the other 10% is without having done the same work yourself is quite the challenge. So the efficiency gains may not be as significant as you think, not until we figure out some smarter processes.

    An example of what I mean: you can feed ChatGPT with notes from an interview conducted with a customer about their experiences and how a product fits into their lives. Supply enough interviews, and ChatGPT can do the work of summarizing them in aggregate, presenting key themes and areas worth looking into, like maybe everyone thinks the price is too high, but it’s because they don’t fully understand the value of what they’re buying.

    It can create a bunch of frameworks to illustrate these findings, like personas and service blueprints. And it can even suggest solutions, like better marketing materials to explain this value to customers. The AI’s output might look pretty good, similar to what a team of human designers would (more slowly) produce, and a company might be tempted to make business decisions based on it. In fact, a team of human designers who haven’t read the interview notes themselves or thought deeply about it might also look at the AI’s work and say it’s good to go.

    The level of confidence and finality inherent in these AI outputs is incredibly convincing. But if a human were to go through all the interviews, listening to the recordings perhaps, they might realize there was a missing element, a feeling subtly unsaid on the margins, that means some customers do see the extra quality, they just wish there was a cheaper version of the product that did less. Skimming through the finished research report crafted by the AI, you wouldn’t even begin to guess where in the sea of correct conclusions this exception could be hiding.

    But there’s no question that this stuff is ready today to do some tasks like image editing, seen in Photoshop’s impressive beta release of a “Generative Fill” feature this week. I took a stock photo and doubled its height, and it was able to get the context of the scene and fill in the missing ceiling almost perfectly. That would have taken an experienced image compositor at least a few minutes, and anyone else way too much time. Just a couple of clicks now.

    I also looked into what Adobe is building with its Sensei suite of AI marketing tools, and that dream of generating and sending personalized ads, as in a unique package of art and copy tailored to a single customer’s behavior, would seem to be already here. I’m not 100% sure how this works everywhere, but in the past, you’d still need copywriters and art people involved in the process after marketers had identified the “customer journeys” and content-goes-here templates. With the opportunities now being identified, advertising messages crafted, and email offers all sent with a single click by the same person, there’s hardly a crack in the door left for the traditional artists and copywriters to make their case. Yet, the quality is good enough to satisfy all but the most discerning of corporations.

    You may observe that the two of the largest advertising platforms are already in this space.

    What do you think about the current advancements in AI and their implications? Share your thoughts in the comments below.

    (One more example: I asked ChatGPT to help suggest edits for this post, and it rewrote one of the above sentences to be better. I liked it, but on closer inspection, there was a glaring semantic error I had to fix myself. It also suggested the call to action above, to increase engagement. Talk to me!)

    ===

    Personal updates

    There seems to be yet another wave of Covid sweeping through the city, based on the fact that several people I know have come down with it, and every bus and train car I’ve been on this week had more people wearing masks, suggesting that they, too, know people who’ve come down with it.

    Kim is going away for a couple of weeks, and I’m hoping she doesn’t run into it out there either; one of her colleagues caught it while traveling in the region a few days ago. I’m planning to stay home as much as I can during this time, and finishing as many video games as possible.

    ===

    Media activity

    • Not a ton of progress in The Legend of Zelda: Breath of the Wild, which I’ve been playing consistently now for the past few weeks — a streak unmatched since the game first came out six years ago (I abandoned it out of fatigue shortly after that initial burst). I’ve now got all four Divine Beasts pointing at the castle and now just need to build up the nerve and arsenal to storm it and be done with this. I seem to be procrastinating instead, exploring areas in the massive world that I never checked out before.
    • The girl band I would say I’m rooting the most for in pop music, XG, performed at the Head in the Clouds festival in New York, and I watched some fancams of their set. The audio quality is terrible in all of them, so I won’t recommend starting there, but they are undeniably polished and tight as a group. Here are two music videos. I think I discovered them back in February, and at the time I said they’re gonna be monstrously huge this year. I stand by this.
    • If you watch the documentary series their label has put on YouTube, you’ll understand why they’re performing at this level: they’ve been physically and psychologically abused for the past six years of training, starting from when some of them were just 12. It’s horrendous to watch, but also probably par for the industry. While it’s good that someone decided to plainly put this footage out there, I’m not seeing much of a backlash, so it’s probably too late and already normalized. Some of the stuff their boss/producer says and does is straight up toxic emotional manipulation (he apparently came up as an idol himself so it’s like Ted Lasso says in the latest episode, hurt people hurt people).
    • Ted Lasso is almost done with its third season, one episode to go. I’m still liking it much better than season two, although it is sooo uneven and odd in its choices. You know the adage, “show, don’t tell”? It’s like the concept of season three is going against that conventional wisdom; a challenge the writing team decided to issue themselves: Can we take lots of scenes that people want to see (scenes of closure, catharsis, and vindication!) and make them happen off-camera and between episodes? And after doing that, can we still make people care through the strength of our set pieces and touching monologues? That’s the only explanation I have for what’s been going on. And to the team’s credit, it works some of the time. It’s not conventional TV, and maybe that’s the point.
    • Platonic, the new sitcommy show on Apple TV+, is much more conventional. It’s about a male and female pair of friends who are really just friends (so far), and a comparison to When Harry Met Sally is drawn in the very first episode. They had a fight and haven’t spoken in years, and then reconnect on the cusp of middle age, when it’s notoriously difficult to form new friendships, let alone platonic ones. I think the concept and set up are strong, but the execution is a little spotty. I’m not really into Seth Rogan’s work, and his character here feels exactly like what you’d expect from one of his characters, but by the end of episode 2 I think I’ll keep watching. The most jarring thing is Rose Byrne’s quasi-Australian accent which raises too many questions about how they met and got along in the past.
    • Speaking of actors whose strong accents shatter the suspension of disbelief, Arnold Fucking Schwarzenegger is back in a NETFLIX TV SERIES which sounds like a dreamy reboot of True Lies. The show is called FUBAR and it’s about a father and daughter who both secretly work for the CIA without knowing about each other’s involvement. I haven’t seen any of it yet, but I’m dying to.
    • It strikes me that in the future, one could give a crazy brief like the above to a generative AI system and start watching something like FUBAR within minutes.
    • My first music discovery of the week is Eternally Yours, a new Alphaville album that sees the band doing symphonic rearrangements of songs such as Big In Japan and Forever Young with a full orchestra. Yes, in Dolby Atmos spatial audio. This is a band that was formed 41 years ago and the lead singer’s voice is still incredible, iconic.
    • The second is Tears can be so soft, the new song by Christine and the Queens. It’s simple but surprisingly soulful, and sonically recalls Massive Attack’s best work.

  • Week 20.23

    • A correction: Last week I said that the the 10th Anniversary release of Daft Punk’s Random Access Memories would have been better if mixed in Dolby Atmos spatial audio. It actually IS! But the entire album isn’t labeled as such because the “second side” of outtakes and behind-the-scenes bits aren’t. With that knowledge, I’ve been enjoying it on headphones and need to put aside more time to hear it again. We regret the error!
    • There’s been a slight Korean bent to the week, in that I’ve had KBBQ and soju twice in four days, watched a lot of Jinny’s Kitchen on Amazon Prime Video, and am actually listening to BTS intentionally for the first time as I type this.
    • Jinny’s Kitchen is a reality TV show about a pop-up restaurant in Mexico run by five of Korea’s most recognizable stars, who go largely unnoticed by their customers. Each of its 10 episodes is 1.5 hours long, which is practically feature film length! They could be much shorter, but it focuses on a lot of mundane, repeated scenes like the staff rolling gimbaps, cooking noodles, and plating fried chicken. Perhaps that’s the point, giving fans a chance to watch them doing normal people things? Or maybe it’s normal for Korean reality TV? Is it designed to be left running in the background? Or was it an Amazon Prime Video decision to hit streaming time KPIs?
    • In any case, Korean soft power is wave after wave of these campaigns. The show should be boring but it kinda works, plus it has an unheard-of 9.2/10 score on IMDB. The best moments for me are when someone does recognize one of them, like when an incredulous customer sees V from BTS (59 million followers on Instagram) pop his head out from the back, and likens it to Drake running a restaurant in a country where no one knows who he is. Watching celebs do normal people things is straight out of the idol playbook and it works — I’m listening to their music with an open mind now, and we’re talking again about visiting the country someday.
    • The Anbernic RG35XX arrived and it’s a nice little device that really recalls the Game Boy series in spirit. It feels good in the hand and the screen is bright and beautiful — the community seems to recommend replacing its software with something called GarlicOS but it looks like wayyy too much trouble. The thing just works out of the box, and I’m not in the mood to start modding it. It still blows my mind you can have this for S$90 including shipping, and it does come with uhh lots of software loaded, so it’s great value if you’re looking for a gift.
    • I’m still going strong in Breath of the Wild on the Nintendo Switch, in the final stretch but honestly still in no shape to face some of the game’s stronger enemies. I’m now thinking that if/when I eventually finish this, I’ll probably have to take a break before starting the sequel, Tears of the Kingdom, given that it reuses the same world map and I won’t be in any mood to cover it all over again. Note to self: don’t pre-order any more games unless they comes with discounts or exclusive bonuses!
    • We had a nail painting team event at work, for which the turnout was more male than female because of scheduling conflicts and maybe even a bit of self selection, so I’m sporting gray/lilac nails at the moment. It’s probably been over 20 years since I’ve had colored nails, and I think I’ll keep this going now that I’m too old to be told off and it’s not such a big deal.

    I asked ChatGPT to analyze my post and it suggested I add a conclusion. I said “you do it, then”. Here’s what we agreed on:

    This week has been a whirlwind of diverse experiences and discoveries. From immersing myself in the captivating sounds of Random Access Memories in Dolby Atmos to exploring the world of Korean soft power through Jinny’s Kitchen, I’ve been on a journey of curiosity and appreciation.

    The Anbernic RG35XX has provided me with nostalgic gaming moments, while progressing in Breath of the Wild has been both thrilling and daunting. Even a simple nail painting team event at work has reminded me of the freedom to embrace self-expression.

    These experiences have sparked my curiosity and inspired me to share my thoughts as a writer. Life is full of opportunities to explore, connect, and find inspiration in unexpected places. So, whether it’s through music, cuisine, gaming, or the little joys, embrace the unknown and discover the stories and experiences that await.

    Thank you for joining me on this adventure, and until next time!

  • Week 19.23

    The new Legend of Zelda game, Tears of the Kingdom, launched this week about five or six years after the last one, which I never finished. I pre-ordered the new game, of course, planning to join the rest of the world on launch day, exploring together and participating in conversations online, collectively figuring out unique solutions using the game’s open-ended physics engine. For those who haven’t seen it, the new game is sort of a sandboxy, Minecrafty affair where you can weld stuff together and build novel mechanical solutions to obstacles, almost certainly in a different manner than your friends. Think rudimentary cars from planks of wood, or hovercrafts, or the forest booby traps from Rambo First Blood.

    But the guilt of never fully playing Breath of the Wild was getting to me, and I’ve been trying to get back into it over the last few weeks. Despite memories to the contrary, I’d made shockingly little progress in my 40+ hours of gameplay, spending most of my time bumbling about the countryside and climbing mountains, instead of conquering the Divine Beasts (1 out of 4) and collecting quality stuff. It seemed wrong to jump ahead to the sequel while I’m finally seeing what the last one had to offer.

    So in this past week I’ve made more progress than in the previous four years: conquered two more Divine Beasts, got the Master Sword at last, and uncovered most of the world map (two more areas to go).

    ===

    Craig Mod tweeted and tooted about having had enough of the iPhone’s (14 Pro, I assume) overprocessed look, and said he was making Halide his default camera app. Huh? But how does that help, I thought, unless he means to shoot in non-ProRAW RAW all the time (which is a thing Halide does: shoot in traditional RAW files which don’t involve the “Photonic Engine” processing pipeline). After some poking about, I realized something I should have ages ago: by turning off “Enable Smartest Processing” in Halide’s settings and choosing HEIC as the output format, you can actually take regular old (non-RAW) photos that look more natural and have more editing latitude! This effectively cancels out the iPhone 14 Pro’s image quality regressions.

    The overstimulated look of the default camera is one of the main reasons I hardly took any phone photos on my recent vacation to Japan, but if only I’d known… I could have! So with that, Halide won an instant annual subscription from me, and I now have a complication on my Lock Screen that launches straight into it.

    My toot about this was boosted by Mod, making it my most “engaging” Mastodon post ever, at 44 likes and 12 boosts.

    You have to see a comparison to understand how obnoxiously oversharpened and artificial the Photonic Engine is on the 14 Pro’s 48mp sensor for some reason. It increases local contrast and highlights texture “detail” that does not look like what you see in real life. It’s a Clarity slider turned all the way up, and I’m not a fan of Clarity sliders. The example below is a corner crop from two larger photos, not edited in any other way.

    Left: iPhone 14 Pro’s default processing | Right: Halide’s “dumb” processing is loads better

    ===

    My AI experiments this week have been around trying to teach ChatGPT to write better poetry. I posted a few days ago about getting it to do a critical reading of a poem, wherein I was completely stunned by its comprehension skills. But writing good poetry, not just gimmicky rhyming lines, that should be a different thing.

    I semi-accidentally discovered that asking a “persona” (that I’d established in another chat instance to test the idea of using an AI prompted with a human backstory as a customer research proxy) to write poetry yielded subjectively better results than when I asked a brand new ChatGPT instance. The mere suggestion that the bot’s point of view should draw references from vaguely described life experiences and preferences seemed to make a difference to what it spat out. Which is why many SF stories have AIs that don’t know they’re AIs, I guess, and they have memories of childhoods that never happened.

    Here’s a poem that “Frank” (quite a salaryman-type personality) wrote about retirement:

    A lifetime spent in labor’s thrall,
    In structured days and office halls.
    Oh, to retire, to end the grind,
    To leave the corporate world behind.

    The dawn of leisure’s quiet grace,
    In this new, unhurried pace.
    Days unbound by ticking clocks,
    Unfettered as the wandering fox.

    Yet in this expanse so vast,
    Shadows of the past are cast.
    Habits of a lifetime past,
    In this new role, steadfast.

    A time of freedom, yet confined,
    By boundaries of the seasoned mind.
    Retirement’s paradox, I find,
    Is having time, yet being time-blind.

    Wow! No? It was coming up with absolute drivel before I handed the job over to Frank in another window. Side note: I started paying for ChatGPT Plus, just to have access to GPT-4; it’s a much better poet than the default GPT 3.5 model.

    So lately I’ve been working on giving feedback to an poetry-writing AI that I named Penrose and instructed to believe in itself as a 50-year-old professor of creative writing and philosophy, who’s bad with money, had two failed “great loves” because of their selfish acts, grew up in Boston, now works in London, and so on.

    Here’s one of the first things it wrote:

    In the chilled reflection of a London morn,
    I see a face, lines etched by yesterday’s turmoil.
    Cracks, not of frailty but life, loud in their silent form,
    That’s you, mate, and it’s all right.

    Witness the march of years, the dance of decades,
    In the ceaseless play of ambition and regret.
    Weaknesses? They’re the worn pennies in your pocket,
    The unpaid bills of your character, yet, not your debt.

    They’re the stumbles on cobblestones,
    The missteps in Milan’s echo-filled alleys.
    Not a damning verdict, but life’s hasty loans,
    A ledger, not of losses, but of rallies.

    In this world where even stone yields to rain,
    We’re all apprentices to time, chiseling away in vain.
    We’re all works, works under the master’s patient hand,
    Unfinished, flawed, yet, in that, something grand.

    ===

    • I haven’t been able to bring myself to watch more of Amazon Prime Video’s Citadel beyond the first two episodes. A colleague said he quit after 10 minutes. It’s reportedly one of the most expensive TV shows ever made, at an average of $50M per episode.
    • Silo on Apple TV+ has been very good so far. It’s a little bit YA sci-fi, a little Hunger Gamey, a little Snowpiercey (but better).
    • I saw a segment on NHK about how this iconic street corner featured in Slam Dunk has been inundated by tourists now that Japan is open again. They interviewed pilgrims from S. Korea, China, and Taiwan who said they just had to come down and see it — “if you love anime at all, this place is a must”. So I decided to get started on the ancient anime, which Netflix has 8 seasons of. The day after seeing episode 1, I ended up standing behind a guy on the train watching the show on his phone.
    • The 10th Anniversary Edition of Daft Punk’s Random Access Memories is out, and the album still holds up extremely well for me. If only they’d come back to remix it in Spatial Audio, that would have been incredible.

  • Week 18.23

    I was in a cab listening to music on my AirPods, and just as we were pulling up, I switched to Transparency Mode and heard a song playing over the car’s radio that sounded kinda familiar. I knew it was a remix of some tune I wanted to know, and managed to Shazam it before getting out.

    Looking into it later, I realized the melody was what I’d been trying to figure out about Charli XCX’s White Mercedes for over a year. Why does that one line she sings — literally the line “like a white Mercedes” — sound like some other song I can’t name? It turns out, it’s literally a song I would have absorbed from the world around me but never intentionally listened to: One Direction’s Night Changes from 2014. Ahhh it’s so good to have that itch scratched! And there are so many more like this I’ve yet to solve.

    Let me say it again for the search engines: Charli XCX’s White Mercedes sounds like, samples, or contains an interpolation from One Direction’s Night Changes.

    Another similar thing happened as I was playing WarioWare Inc. (GBA, 2003) again for the first time in years. The background music in one stage awoke some long dormant memory and I needed to know what pop song from my younger days it sounded like. After a lot of humming aloud and trying to Shazam it and searching online… I concluded that the song it reminded me of was… itself. It’s called Drifting Away, and I must have really loved it back when I was playing the game for the first time.

    Speaking of retro games, I lasted a full week. The Anbernic RG35XX I said I wouldn’t buy since I already have a Retroid Pocket Flip is now on the way to me from China. There are some reports of shoddy QA and long-term durability, but for S$90 I think that’s to be expected.

    ===

    Another week, another bunch of water-cooler conversations about AI. Specifically how it relates to our work in design: as accelerator, collaborator, ambiguous combatant, amoral replacement. I don’t just mean the making pictures and writing words part, but analyzing messy human interactions (it’s just unstructured data) and presenting them in new ways.

    I ran one experiment with ChatGPT on Sunday afternoon, just for kicks, and it sort of blew my mind. From a handful of behavioral traits and demographic details I supplied, it was able to inhabit a fictional personality that I could speak to and pitch various products to. So far so par for the course. But then it reacted to a hypothetical KFC offering called “The Colonel’s Colossal Combo” in a way I didn’t expect, citing a conflict with values and dietary preferences that I did not specify. When asked where they came from, it argued that although they were not specified, they could be reasonably expected from the “Frank” persona I’d created, because of some other background that I DID provide. It sounded a lot like intelligent reasoning to me, and regardless of how it works, I was happy to accept the inference the same as if a colleague were making it.

    Like with all advances in automation, it’s inevitable that we’ll now be able to (have to) do more in less time, with fewer people. Until things go “too far” and need to be reined in, it’s not even a question of whether we should — every industry is incentivized to discover when can be done before it gets done to them. I think there are some exciting opportunities for designers, and a slew of unknown consequences for society. And just like that, we’re back in a new “fuck around” phase of the tech cycle.

    ===

    A couple of weeks ago I made a bunch of fashion-style athleisure photos with Midjourney v5 but somehow forgot to post them. The photorealistic ones are quite incredible, and the few illustrations I got were really strong too.

    Midjourney v5

    This week, v5.1 dropped, promising more opinionated outputs and sharper details, so I tried the same prompt again. Many of the results were as broken as these bodies.

    Midjourney v5.1

    They probably fixed something quietly because it’s been more reliable in the days since. I thought it would be interesting to do a comparison of models 1 through 5.1 with the same prompt. It’s crazy how far it’s come in just over a year.

    photograph of Queen Elizabeth II in a dim video arcade, sitting at a street fighter 2 arcade cabinet, intense concentration playing game, side view, screen glow reflected on her face, atmospheric dramatic lighting --ar 3:2
    v1: Feb 2022
    v2: Apr 2022
    v3: Jul 2022
    v4: Nov 2022
    v5: Mar 2023
    v5.1: May 2023
    v5.1: very high style variation
    v5.1: raw mode + 2x quality

    If you saw Midjourney a year ago, you were probably impressed by how it and Dall-E 2 could turn quite natural text descriptions into imagery, even if the results were still quite hallucinatory, like DeepDream’s outputs circa 2015. I don’t think you would have expected to see the pace of improvement be this quick.

    It’s not just rendering improvements from distorted pastiches to photorealistic scenes with internal logic (global light affecting nearby objects realistically, fabrics folding, leather seat covers stretching under buttocks), but also how it’s evolved through feedback and training to understand intent: the idea of a “side view” started working from v4. None of the earlier re-generations got me the camera angle I was going for. The tools that promise to do this for video are probably going to get good faster than you expect.

  • Week 17.23

    • I discovered that Midjourney has an alternate set of models called Niji (aka Nijijourney) dedicated to creating anime-styled imagery. It’s astoundingly good. It has four stylistic modifiers: standard, cute, expressive, and scenic. Look at all the implicit context and environmental storytelling in these scenes. I really wonder where they came from.
    An izakaya where bears and humans drink together
    “Cute” version
    This was meant to be Singapore but I don’t see any evidence of it
    • I also found Draw Things on the App Store, for both iOS and macOS, which can download an array of open source AI image generation models off the internet and run them locally on your devices — no fees, no internet connection required. Grab it while you can. Of course they are nowhere as advanced or fast as the paid services, but you know they’re going to get there soon, especially if Apple continues to crank up their proprietary silicon. Incidentally the anime-focused version of Stable Diffusion is called Waifu Diffusion.
    • My Retroid Pocket Flip arrived from China and I was relieved to find it quite a solid product. The build quality is good, no looseness or wobbles; the D-pad and all buttons feel great; the screen is incredibly bright; and the giant 5,000mah battery and active cooling make it more than just an Android phone with physical controls attached. It’s a really nice way to run emulated ROMs. I used to love playing Lumines and Every Extend Extra on my PSP, like over 15 years ago, and being able to revisit them again on this little $164 USD device is quite a thrill.
    Processed with VSCO with kc25 preset
    Processed with VSCO with kc25 preset
    • If I hadn’t impulsively pre-ordered this while in Japan, literally while walking to our anniversary dinner in west Shinjuku, then I would definitely be buying an Anbernic RG35XX right now for a mind-blowing $56 USD. It’s a Game Boy Pocket-inspired device with a bright 640×480 screen and the ability to emulate all 32-bit consoles, and maybe even the N64. I can’t believe how cheap and good these things have gotten, and there are so many of them on the market too.
    • I finally finished watching the Korean revenge drama series The Glory — it took awhile because Kim wasn’t interested and so I only get to see it on my own time. It’s the rare TV show that dares to wrap up its core story in the first season, and The Glory gets some very satisfying closure in. The remaining threads could make for an interesting second season (now in production), but also it could have been canceled and everyone would be okay.
    • We started watching Drops of God on Apple TV+, which starts off with an immediate deviation from its manga source material about the world of wine: a Japanese male main character has been replaced by a French female one. This adaptation is an international joint production that switches between English, French, and Japanese, and each episode begins with a reminder ‘not to adjust your television’. Anyway, I think they’ve managed to keep the main idea while toning down the big, overdramatic ah hah! moments you’d expect from manga/anime. It still has people honing and demonstrating their near-superhuman skills (taste and smell, in this case), which is always fun, even when said skills aren’t the usual martial arts, boxing, tennis, math, golf, you get the idea.
    • Everything But The Girl’s comeback album, Fuse, is officially a hit. It debuted at #3 on the UK charts, a stunning career best for Tracey Thorn and Ben Watt. It’s so good to see musicians from <wheezing> our generation </wheezing> coming back after a long hiatus to demonstrate absolute mastery of their craft (as opposed to embarrassing themselves, e.g. U2, The Smashing Pumpkins).
    • Michael also mentioned the greatness of Karma Police out of nowhere, which led me to play the song in my head, and I commented that it simply sounds like nothing else. I tried asking Apple Music to make a radio station from similar songs but it was totally wrong, just songs from bands in the same wide category, but none of them actually sharing the same vibe or brilliance. Somehow this led to me revisiting Keane’s very strong first album, which I have many strong emotional attachments to. It came out in 2004, I think, and I must have played the hell out of it.

  • Week 16.23

    I usually look through my camera roll to recall events as I start writing these posts. It’s telling me nothing much happened this week.

    That’s not true; it’s just a lot of it was spent online. You might have noticed the excitement and fast pace of advancements in AI recently, and it seems I’m spending a correspondingly larger amount of time playing with, reading about, and discussing the impact of it on our work and lives. It’s enough to make one consider taking a gap quarter or year off work to focus on this stuff.

    One catalyst was a colleague being invited to do an interview on what it means for design, and so we had a conversation about the trends beforehand. Unsurprisingly, the media is still thinking about both design and AI simplistically: will image generation mean fewer jobs for illustrators and that sort of thing. I find it hard to be optimistic in the short-term, in that AI is lighting a fire under our asses and it’s going to cause a lot of pain. But the potential for us as a discipline to evolve under pressure into something greater is undeniable.

    It didn’t help that the next thing I saw was The AI Dilemma, a talk by the creators of the documentary, The Social Dilemma, wherein they say the problems unleashed on society by social media were just the prequel to what AI is on track to do if we don’t prepare. And let’s just admit we don’t have a great track record of preparing for things we know are going to hit us later. It’s about an hour long but I’d file it under essential viewing just for awareness of what’s building up.

    The above talk was given at The Center for Humane Technology, and coincidentally this was the week we finally got a look at what Humane, the secretive product company founded by a load of ex-Apple designers and engineers, has been building and teasing.

    I’ve been anticipating their debut for a long time and had a pretty good idea of the core concept from their leaked pitch deck and patents. Essentially, a device achieves AR by projecting a digital interface on the world around you the old-fashioned way, using rays of light pointed outwards, rather than on the inside of glasses. At some point along the way they started mentioning AI a lot, and it looks like the secret ingredient that turns a nothing-new wearable camera + laser projector into a real alternative to smartphones. In other words, an intelligent assistant that isn’t primarily screen based, so we can be less distracted from “real life”.

    It’s probably best to withhold judgment until we see more at some sort of unveiling event, with more demos, a name, a price, a positioning. But it’s worth remembering that when the iPhone came out, it was a phone good enough to replace whatever you were using at the time. Humane’s device is said to be standalone and not an accessory to be paired with a smartphone. It’s also shown taking calls. The bar for replacing your telephone is now much higher after some 16 years of iPhones.

    An intelligent assistant that let you do things quicker with less fiddling was always my hope for the Apple Watch from its very first version; that Siri would be the heart of the experience, and the UI wouldn’t be a mess of tiny app icons and widgets, but a flexible and dynamic stream of intelligently surfaced info and prompts. We all know Siri (as a catch-all brand/name for Apple AI) wasn’t up to the task at the time, but I keep hoping the day is right around the corner. Fingers crossed for the rumored watchOS revamp at WWDC this year.

    There’s now also a rumor that iOS 17 will add a new journaling app, and my expectations are already very high. They say it’ll be private, but tap into on-device data like Health and your contacts and calendars. That goes beyond what Day One does. I’m imagining the ultimate lifelogging app that automatically records where you go, who you met, what you did, how tired you were, what music you were listening to, and your personal reflections, all in one searchable place. I’ve tried a bunch of these before, like Moves and Momento, but nothing lasted. If Apple does do this, I may finally be able to ditch Foursquare/Swarm, which I still reluctantly use to have a record of where I’ve been. Its social network aspect is nice but not essential since hardly anyone else uses it now.

    I remember there was a Twitter-like app called Jaiku on Nokia smartphones over 15 years ago that had a feature where, using Bluetooth, it could tell if you met up with a fellow user, and post to your other friends about it. I was excited by it but had few friends and even fewer ones on Jaiku. Just like with AirTags and Find My, tapping into Apple’s giant user base could finally make this concept viable. As long as Apple isn’t trying to do a social network again.

    ===

    Oh right, back to AI. What have I been doing? Some of it was playing games with ChatGPT, essentially asking it to be a dungeon master using the following superprompt (which I did not create btw!):

    I want you to act like you are simulating a Multi-User Dungeon (MUD). Subsequent commands should be interpreted as being sent to the MUD. The MUD should allow me to navigate the world, interact with the world, observe the world, and interact with both NPCs and (simulated) player characters. I should be able to pick up objects, use objects, carry an inventory, and also say arbitrary things to any other players. You should simulate the occasional player character coming through, as though this was a person connected online. There should be a goal and a purpose to the MUD. The storyline of the MUD should be affected by my actions but can also progress on its own in between commands. I can also type “.” if I just want the simulated MUD to progress further without without any actions. The MUD should offer a list of commands that can be viewed via ‘help’. Before we begin, please just acknowledge you understand the request and then I will send one more message describing the environment for the MUD (the context, plot, character I am playing, etc.) After that, please respond by simulating the spawn-in event in the MUD for the player.

    Try it! I even had success asking it (in a separate chat) to come up with novel scenarios for a SF text adventure game, which I then fed back into this prompt. I can’t emphasize enough how fun this is: you can take virtually any interesting, dramatic scenario and immediately play it out as an interactive story.

    Here’s an example where I played the role of a time traveler who has to stop a future AI from destroying humanity by going back in time to prevent the invention of certain things, starting with the Great Pyramid of Giza, which will purportedly become a power source for the AI.

    And here are a couple of new products made possible by GPT. There are so many, all asking for about $10/mo. Most won’t survive as this stuff becomes commoditized, but for the moment they are all amazing because these things weren’t possible before.

    • Tome: It’s a sort of PowerPoint that can create entire decks on its own from a short brief you give it. For example, ask for a sales deck and it’ll set up a working narrative arc over multiple slides, not filled with placeholder text and images mind you! But actually generate text and original pictures to fill every one of them. Of course, it will use common storytelling structures — the portfolio introduction I made as a test looked like 90% of the applications that we see, using very familiar language for describing one’s experience, design philosophy, values, skills. This is fine, of course. You can edit it, or use it for as long as “what went before” continues to have currency in this society. When quality is everywhere, quality becomes meaningless. Fire under buttocks.
    • Rationale AI: Describe a decision you’re trying to make, and it’ll tell you the pros and cons, or generate a SWOT analysis, or work out the causal chain of the path you’re on. For many people, this sort of reasoning is not hard to do, but perhaps it’s a game changer for those who can’t. For example, if you’re in an emotionally distressing situation and cool logic is evasive; it could help to show the bigger picture. I tested it with such a scenario and it gave some solid insights (be careful with advice from an AI, of course). But that this thing works at all is a marvel! “Should I become a full-time influencer?” is not a question a machine could have understood in the past, and certainly it could not have forecasted that failing down the road might put stress on your finances and lead to harmful self doubt and regret over quitting your job.
    • Summarize.tech: I found this by accident when someone shared a two-hour YouTube video essay in a group chat and everyone said “I ain’t got time for that”. I remarked that it sure would be great if an AI could watch that and write a tl;dr for us. And then I thought… surely that exists. And it does.

    ===

    It was also my birthday, and I saw John Wick 4 and ate a lot of Taiwanese hot pot. Also binged all of the new Netflix show, The Diplomat, and it was actually good. Life’s alright when that happens.

  • Week 14.23

    Ugh, the post-holiday period is the worst. I’ve struggled through the week, and it was only a short four-day work week because of the Easter/Good Friday holiday. I’m in the mood for another break now, and thankfully we have a week in Australia later this year to look forward to.

    I started off Monday with a client video call in which I got frustrated enough by my bad lighting situation (sitting in front of blinds — either too much light, too little, or visible horizontal shadows across my face) to finally do something about it. During my aimless ambles down the aisles of Japan’s electronic superstores, I saw many shelves dedicated to remote work equipment, presumably a big sales driver for them over Covid-19, and considered bringing a ring light home. I didn’t, but I found good looking ones on Shopee and ended up with a rectangular soft LED panel on a tabletop stand for just S$27! It does five color temperatures, but I’m sticking with Daylight, and overall it’s been an awesome purchase I should have made ages ago. And it arrived in 24 hours.

    No surprises, but I’ve taken far fewer photos since returning. I still open Hipstamatic regularly just to keep my streak going, and it’s forced me to try and snap something every day. That said, I wonder if this habit, and the product’s reboot, will last. As I was discussing with Michael, they needed to put some momentum behind the launch and sustain it with updates and quality posts in the global pool. But from how it looked in their updates, the founders were (also) on holiday in Japan on launch week? Perhaps they were there to boost some community events, but I looked at the Japan-only photo feed regularly and I was one of the most prolific posters. Not a great sign. They just released an update this weekend, at least, with a new Uji-inspired lens and film.

    A new fun thing to do with Midjourney emerged this week: a /describe command which takes a photo you upload and has the system describe it back to you in the form of Midjourney prompts, which you can then submit to generate a “broken telephone” remix of your original image.

    If you think computer vision/image recognition has gotten scarily good recently, you’d be right. AI is part of this chain somewhere, and look no further than this Memecam web app which blew my mind last night. Snap a photo of something, and it recognizes what the image contains, and uses GPT to create a joke and final meme, Impact font and all. It actually writes jokes about anything, instantly. That AI-generated Seinfeld stream could technically become good, viable (if not wholly original) comedy in the near future.

    ===

    Hey, two quick moments of consumer ecstasy I need to share!

    • We’ve got the HomePods in Singapore at long last. I don’t know what took Apple so long, but you can now officially buy them here, and the prices are slightly lower than I would have expected, at S$139 and S$429 for the mini and full-sized second-generation HomePods respectively. My Sonos speakers are now unplugged and we are a fully Siri-ed home. I’d previously bought two minis for the office and bedroom off the gray market, and those are now joined by two large ones in the living and dining areas. Reader, they sound glorious. It’s a rich, tangible, and emotionally satisfying experience for your favorite music. There was a point in time where Apple loved the word “magical” and used it liberally. Even for mundane things like keyboards that worked reliably. But these, these are kinda magical.
    • Nespresso launched a new kind of pod locally, one designed to approximate “filter-style” coffee, which in my mind is basically a pour over. They’ve been out for a few months in limited European markets, it seems, but still not widely available. They have a new design where you peel off a sticker to reveal an in-set dot grid which the liquid passes through — the foil is not punctured as a a result. You’re meant to press both the Lungo and Espresso buttons in sequence, resulting in a 150ml extraction, which they call a Gran Lungo. Lol. Anyway, it tastes pretty good. The longer cup is thinner and more delicate than if you used a regular pod to do an Americano or long black, with hardly any crema. This innovation allows for floral and fruity roasts to come through, if you like that sort of thing. I..I..I think it also results in more caffeine.
    • Boss coffee is now natively available in Singapore! Used to be you’d find imported cans in Don Don Donki (the local name for Don Quijote) and some other Japanese supermarkets, but it looks like Suntory is properly trying to bring “The No. 1 Ready To Drink Coffee Brand in Japan” to Singapore now. But I remain unconvinced these stubby plastic bottles and generic labels are the way. The little cans and their designs are iconic, and better for the environment since they can be recycled.
    Spotted in a 7-Eleven

    Last week I mentioned buying black tees from FamilyMart, and then got into a few brief discussions about fashion/luxury apparel this week, wherein I reflected that while I’m happy to pay high prices for technology and things crafted out of metal, I can’t feel that way about fabrics and leather. They wear down, so why not just embrace their replacement and buy cost-effective, expendable products from basic brands? Then the Twitter algorithm put a bit of trivia in front of me that the plain white tees worn by Carmy in The Bear got attention from viewers who wanted to buy them, and that they were actually pretty expensive ones made by Japanese brand, Whitesville.

    So… if you know me, you may know where this is going. Yup, this is the guy who loved PCs, hated Macs, and now has a house full of Apple products. To be clear, I wasn’t suddenly curious about the idea of buying ostentatious Veblen t-shirts with designer logos, just… better ones that would hold up longer and not look as cheap. So I now have an order of basic black tees coming in from Mr Porter that cost 5x what I normally pay for them. Gulp. I’ll work out if this actually makes sense and let you know.

    ===

    The Super Mario Bros. Movie is a fun trip in IMAX. We enjoyed it, and I’m looking forward to finally playing Super Mario 3D World + Bowser’s Fury and New Super Mario Bros. U Deluxe (I have to look up these names every single time) on my Switch soon.