When Design Defeats AI

You're receiving this newsletter because you showed interest in Productize Philosophy, an Interaction Design newsletter that introduces a new way for understanding and creating products emerging from philosophical practices and notions from the arts of persuasion. If, instead you received this as forward, you can subscribe here for all the goodness. Happy Reading!

Today's key takeaway: "What modern product/interaction designers must take responsibility for is: creating more Latent Centaurs in our world."

‍

Hey, Everyone.

I was gone for I don't know how long. I hit my first major creative rut and decided to make a trip to my hometown after about 9 months. It took me a little while to get back to the flow of writing, and I ended up writing 4-5 distinct drafts before I settled on today's discourse. Let's begin right away.

‍

🎙 Today's Discourse:

When Design defeats AI

The outcomes of AI are unpredictable. Which can make a system run by AI an unpredictable experience. Such is the case for everyone of you who have lived within GMail’s client. There is a degree of uncertainty as to what will arrive in the Inbox, and what will get distributed across Updates, Promotions, Social and even, Spam (I once had a reply to my reply end up in spam).

Once when I decided to purge all of these categories and have only Inbox, the mess piled up, I got tired of blocking while the uncertainty ensued as to who can even be sending emails to my Inbox in the first place. This is the very problem that Google sort to solve with an AI distributing emails across categories. A few years ago, Google had even gone as far as creating a whole new product called Inbox by Google, where everything was supposed to be “smartly” segregated and presented to you if and when you need it. All your email needs presumably, automatically pegged by the AI. Of course, there is no such AI that can reliably hit that ideal and Google eventually killed the product altogether.

In a bid to help you clearly manage your Inbox, dozens of companies have created many clients to run over GMail, promising to provide you with a fast, highly efficient emailing. But none quite succeeds at removing the uncertainty pervasive to your Inbox.

Enter in the chat: HEY E-Mail.

The first thing HEY does is let you filter out who can send emails to your Inbox. The first time someone sends you an email, you can read the email and choose to either screen them in or block them out for eternity. This system alone works really well (even the “for eternity” part). In most situations you can actually be really sure if somebody would be making good use of your email attention. But even when you’re not, HEY has the provision to let you simply clean out the screener stack without any blocking and decide the next time they send you an email.

The design lets no one’s email to land in your Inbox without your explicit permission.

If the HEY screener was only about filtering emails through to your Inbox, it would have failed. But what HEY further does is let you also pick if the incoming emails will be of the Save-only type or Read-only type. The former goes to a different container titled “Paper Trail” and can include emails like receipts, the latter goes to “The Feed” and can include your newsletters. And this leaves all the emails that you need to interact with and threads you need to stay on top-off to arrive in Inbox, or as HEY’s designers decided to label it as Imbox.

It has been exactly a year since I have been using HEY and lo-and-behold, this system works. I am very aware of what lives in my emails, and appreciate knowing the kind of attention each container will demand. There are additionally Read-Later stack [called “Set Aside” in HEY] and Reply-Later stack that you can push emails to while skimming, which ensures you don’t have to skim through your emails and act on them at the same time. This has creative value as composing an email is seldom the same cognitive process as triaging through your emails. I feel in control of my emails for the first time since ever.

…without needing an AI

Design Solutions v/s AI Solutions

When HEY came out, I deconstructed the entire product’s design in an essay I am quite proud of. I write about HEY again today for I can’t help but wonder what goes on in the product team of GMail. Why does nobody think to re-ask the design questions that concern email. You are trying to create a service for person A to communicate with person B. Start with this premise and build your away up [as I show in the HEY essay]. But rather embarrassingly, GMail is Electronic Mail in the image of Physical Mail: Separate boxes for Incoming and Sent, and anybody can drop anything in your Inbox. WHY do we have the same structural design for an entirely different technology!

Google never innovated on any structural design aspects of email. Instead, they assumed an AI can manage all human communications.

But human communications begins with the medium (i.e. structural design) before it concerns itself with the message. The medium holds the communications together and any problem-solving must start there. Yet, this is the very idea that is lost on all Big Tech, as they keep throwing AI to manage Inboxes, Feeds, etc., managing at the message level, much to the frustration of their users who are left feeling that yet another entity, an AI Algorithm, doesn’t understand them and they don’t understand it. Uncertainty ensue.

The simplicity of alternative design solutions will surprise you. For instance, Instagram can solve most of user’s frustration if it merely bifurcates the feed into two: one for Public and another for Private accounts, as these are two different structures producing two different kinds of posts [for me, it will translate to separate feeds for Friends and Creators]. Today, due to my DM activity, Instagram’s AI constantly prioritizes my Friend-posts at the expense of Creator-posts to the point I now rarely see their art appear on the feed.

(I have now manually muted all my friend posts to get back to seeing artists in my insta-feed again.)

AI can't reach Creativity

Human expression and communications is a creative endeavor, and as is true with any creative endeavor, AI can't understand it. Many argue that AI will eventually get better over time, but here I will intercept and say, “It won’t.” Human Creativity is not something you can hope to understand with more data or better algorithms. Both are limited by the very medium they exist in. The creativity phenomena lies beyond the language and the numbers, in the latent space that exists between our models. Latent thinking is for the human. AI, by nature of it’s technology, can’t reach it.

To all product creators, the right way then is to

• create designs by identifying where human creativity exists in a flow,
• design structures to capture every creative intention, and then
• bring technologies like AI to help in a way that is legible to the user.*

*Legibility is not to be confused with making all the working details visible in Design. Legibility concerns with predictable outputs.

The makers of HEY neatly writes about this in their manifesto:

"Email’s better with a human at the helm. That’s you. You’re better at deciding where things go, what your intentions are, and how you want things set up. The machines have a lot of learning to do before they'll be able to second-guess whether you actually wanted to see that email, whether it was a receipt or a newsletter, and even what you should be writing someone. At HEY, it's human intelligence over artificial intelligence. HI > AI."

AI-augmented Humans

Peter Thiel in his book Zero to One writes,

“The most valuable companies of coming decades will be built by entrepreneurs who seek to empower people rather than replace them.”

We must strive to augment Human Intelligence with Artificial Intelligence, instead of the latter replacing the former (as is popularly the case with content recommendations and other social media phenomena today). Elon Musk remarks this replacement as the real concern with AI. When public discourse of the entire planet bends at the whims of an opaque algorithm controlled by few engineers in a room, it is a serious cause for concern.

Better Design can replace the need for AI and constructively deploy AI where useful. Design questions must help us figure out where AI is needed.

GMail’s auto-complete is meant to be seen as an AI meant to help us be better email writers. And you know what: *it actually can.* There is nothing wrong with having alternate suggestions to what you’re writing, however, the way GMail has designed around it is critically intrusive to creativity. The suggestions appear right beside the cursor begging to be used. Instead, it should be presented as an option *silently* waiting for you to click and check.

Today, the way GMail’s AI suggestions are presented feels imposing to creative thought train.

Design for AI augmenting HI. In a recent essay called Superhistory, Not Superintelligence, Venkatesh G Rao gives, perhaps, my favorite framework for AI so far. He explains how Magnus Carlsen is the first world champion trained by AI. The key insight here is that AI can play way more games in an hour than a human would in a lifetime. This doesn't make AI "super-intelligent" but "super-historic" in it's digestion of information worth many lifetimes. When Carlsen dethroned Viswanathan Anand in 2013, it marked a tectonic shift in how chess games progress at the highest levels. His plays are more AI inspired (exploring longer games and more possibilities) than the Humans who played before him. He is, as VGR calls him, a Latent Centaur. After reading VGR's essay, I realized that what modern product/interaction designers must take responsibility for is: creating more Latent Centaurs in our world.

​

​

📸 Today's Snapshot Moment

To all product creators, create designs by identifying where human creativity exists in a flow, design structures to capture every creative intention, and then bring technologies like AI to help in a way that is legible to the user.*

‍

📚 Curations

"... to get to the post-textual, post-verbal era, when words are for AIs and humans have to think in post-latent spaces where the AIs can’t reach." - @vgr​

Design for AI augmenting HI. In a recent essay, Venkatesh G Rao explains how Magnus Carlsen is the best chess player in the world and is trained by AI. His plays are more AI inspired than the Humans who played before him. He is, as VGR calls him, a Latent Centaur. What Product designers must take responsibility for is creating more Latent Centaurs in our world.

"...And it has led me to start thinking about AI in terms of time rather than intelligence."

(AI is) Superhistory, not Superintelligence *MUST READ*

​Algorithmic Blindspots by David Perell

​How to Escape Social Media Addiction (and Algorithms) by Me

‍

🏷 Almost Too Short Reads

1. Nothing Ear(1) is definitively better than AirPods!

First up, I love the design aesthetics. The aesthetics can carry the earphone on its own. It is one of those designs that is really expressive, had a character, rather than being muted, "sleek", touted as elegant, minimal but is rather utilitarian. What I love most about Carl Pei [founder, Nothing and before this, OnePlus] and the collaborating team at Teenage Engineering is that they went with a clear casing. A clear casing is incredibly hard to achieve as it requires you to also focus on how the internal circuitry looks and ensure that all the glue have to be precisely applied to look clean.

​

Functionally, Nothing earphones hits the threshold of noise-cancelling by being effective in airplanes [as MKBHD reports]. To my liking, the sound profile is flat and balance with no bass-boosting out of the box. Priced at $99 where no one even offers noise-cancelling, buy the Nothing Ear(1) if you're not willing to spend more (and like 60% more for the feature set). I see not much reason AirPods anymore.

If you're interested in spending more, Sony released their latest earbuds which blows the competition out of the water and achieves a new height for wireless earbuds: a noise cancelling almost as good as headphones. Check it out here!​

2. Dark Mode is Vanity Design.

I never really got into dark mode trend, and never really got excited when my favorite apps got dark mode. Whenever I showed my lack of enthusiasm for the aesthetic, people reverted back with functional reasons like easy on the eyes and so on. But... there is a palette of colors to be explored that satisfies that need. For ambient aesthetics, we can simultaneously play across Tone : muted, saturated; Texture : matte, glassy, and Illumination : physical light gradients to achieve easy-on-the-eyes-in-the-dark need. Kindle's and Deepstash's beige aesthetics are few simple examples but there is room trying so much more.

Most dark mode implementations is a simple interchanging of colors that imho looks very out of place, especially noticeable when you've just switched from light mode. Consider the pioneer of dark aesthetics, Spotify, where dark-mode feels like the native nature of it's UI. Spotify have made dark-mode specific choices like illuminating the background for darker environments, a sort of glossy texture for buttons, a smaller, finer font suitable for white-on-black text, and many more. These are dark-mode native design decisions which makes the UI feel right. Simply switching to dark mode by switching colors is a hacky-job, and touting it as an major design feature is a vanity chase. Dark mode is vanity design.

Done!

I think it has almost been a month since I sent you a newsletter, and I had hit a bit of a creative rut. I would slip into careless opinions every time I sat down to write, usually diverging from whatever I was writing about. Nothing about it felt right. I had gone back to my hometown during this time and I realized my workflows are not consolidated enough to allow me to work wherever I wish. The questions of remote work!

I am glad I returned today with a fairly content-heavy letter. It feels a major win to have finally finished a long-one.

Help me out here, can you? If you know someone you think can benefit from these newsletters, please reply back and I will personally mail them along with the best ideas of Productize Philosophy so far. I promise to make you look good.... unless for some wonky reason 🤣 you don't want me to mention you.

Support me by giving me a shoutout on Social Media.

​Share this NewsLetter on Twitter | LinkedIn | Facebook | Pinterest | Mail to Someone​

And, ofcourse, if you are new here, first of all, thank you so very much and here's the PP archives for you to binge. Cheers! 🍻 Let me know how it goes:

Productize Philosophy Archives

​And as always, do reply with any thoughts, criticisms and feedback. It is of most value to me.

​
Until Next Week *sincerely this time*,
Abhishek Agarwal
​Blog | Twitter‍

PS. Subscription Form at the Bottom of this Page.