This week was an exciting one for the AI ​​community, as Apple joined Google, OpenAI, Anthropic, Meta and others in a long-running competition to find an icon that even remotely suggests AI to users. And like everyone else, Apple threw down the gauntlet.

Apple intelligence is represented by a circular shape composed of seven loops. Or possibly it is a circle with a crooked infinity symbol inside? No, it’s New Siri from Apple Intelligence. Or possibly it’s New Siri when your phone lights up around the edges? Yes.

The thing is, nobody knows what artificial intelligence looks like and even what it should seem like. It does every part but looks like nothing. However, it needs to be represented in user interfaces so that people know they’re interacting with a machine learning model and not only an easy search, submit, or whatever.

While approaches to stigmatizing this supposedly all-seeing, all-knowing, all-acting intelligence vary, they’ve coalesced around the idea that an artificial intelligence avatar must be non-threatening, abstract, but relatively easy and non-anthropomorphic. (They seem to have rejected my suggestion that these models at all times speak in rhyme.)

Early icons of artificial intelligence were sometimes little robots, wizard hats or magic wands: novelties. But the implications of the former are inhumanity, rigidity and limitation – robots know nothing, usually are not personal to you, perform predetermined, automated tasks. And magic wands and the like suggest irrational invention, inexplicable, mysterious – perhaps advantageous for an image generator or creative soundboard, but not for the sorts of actual, credible answers these firms want you to consider that A.I. ensure.

Company logo design is generally an odd mixture of strong vision, industrial necessity, and compromise made by committee. The effects of those influences could be seen in the logos featured here.

The strongest vision is, for higher or for worse, the black dot of OpenAI. The cold, featureless hole into which you drop your query is a bit like a wishing well or Echo’s cave.

The committee’s best energy was directed towards Microsoft, whose Copilot logo is virtually indescribable.

But notice how 4 of the six (five of the seven, when you count Apple twice and why we shouldn’t) use nice, candy-colored colours: colours that mean nothing but are cheerful and approachable, leaning toward femininity (how such things are considered in design language) and even childish. Soft transitions of pink, purple and turquoise; pastels, not hard colours; 4 are soft, limitless shapes; Embarrassment and Google have sharp edges, but the former suggests an limitless book, while the latter is a pleased, symmetrical star with friendly concavities. Some also animate in use, feeling alive and responsive (and crowd pleasing, so you possibly can’t ignore it – taking a look at you, Meta).

Generally speaking, the intended impression is one among friendliness, openness, and undefined potential – as opposed to points comparable to expertise, efficiency, decisiveness, or creativity, for instance.

Do you think that I’m analyzing an excessive amount of? How many pages do you think that the design development documents for every of those logos contained – greater than or lower than 20 pages? My money can be on the former. Companies are obsessive about these things. (And yet you by some means missed a logo of hate or created an inexplicably sexual atmosphere.)

However, the point is not that corporate design teams do what they do, but that nobody has managed to give you a visible concept that clearly says “AI” to the user. At best, these colourful shapes convey a negative idea: that this interface is an email, a search engine, a notes app.

Email logos are sometimes related to an envelope because (after all) it is email, each conceptually and practically. A more general “send” message icon is indicated, sometimes split, like a paper plane, indicating a document in motion. The settings use a gear or key, suggesting tinkering with the engine or machine. These concepts apply across languages ​​and (to some extent) generations.

Not every icon may refer so clearly to its corresponding function. For example, how do you indicate “download” when the word varies from culture to culture? In France they charge a telécharge, which makes sense, but doesn’t really mean “collecting”. However, we have now reached the arrow pointing down, sometimes touching the surface. Load. The same goes for cloud computing – we have embraced the cloud although it’s mainly a marketing term meaning “a big data center somewhere.” But what was the alternative? Little data center button?

AI is still recent to consumers, who’re being asked to use it as a substitute of “other things,” which is a really general category that AI product providers are reluctant to define because that would mean there are some things AI can do and a few things it will probably’t do. could be. They usually are not ready to admit it: all fiction is based on the fact that artificial intelligence is able to do every part in theory, and achieving it is only a matter of engineering and computation.

In other words, to paraphrase Steinbeck: every AI considers itself temporarily embarrassed by AGI. (Or I should say, it is considered by the marketing department, because artificial intelligence itself, as a pattern generator, doesn’t take anything into consideration.)

In the meantime, these firms still have to call it what it is and put a “face” on it – though it’s telling and refreshing that nobody has actually chosen a face. But even here they’re acting at the whim of consumers who ignore GPT version numbers as an oddity and like to say ChatGPT; who is unable to establish contact with “Bard”, but agrees to meet “Gemini”, who has proven focus; who’ve never wanted to Bing (and definitely not talk to the thing) but don’t mind having a co-pilot.

Apple, for its part, has taken a shotgun approach: you ask Siri to send a request to Apple Intelligence (two different logos), which takes place in your private cloud (unrelated to iCloud), and possibly even forward your request to ChatGPT (no logo allowed) ), and the best indication that the AI ​​is listening to what you are saying is… swirling colours somewhere or in every single place on the screen.

Until artificial intelligence is defined just a little higher, we will expect the icons and logos that represent it to proceed to be vague, non-threatening, abstract shapes. A colourful, ever-changing blob would not take your job, right?