In celebration of spelling mistokes, cats on keyboards, and the things which make us human

AI is eating itself, and we're only here to prop up the share prices. It's time to value not just your thinking, but your foibles too - it's all that separates us from the bots.

by Rich Fairbairn

Published: 17:52, 02 January 2025

Hello, my name is Richard, and I am a chronic typo-ist.

My keyboard (it’s not me, definitely) leads me into making more spelling mistakes per paragraph than the proverbial cat on a keyboard. I leave it to others to judge which of us makes more sense.

Over the eons I’ve been at it, I’ve got used to building in extra time for every deadline, just so I can go back over each sentence and fix up the trail of mistakes which follow the cursor like Hansel & Gretel marking their escape route from the forest.

Being both frugal and resistant to needless shopping, I continue to put off changing a perfectly functional sleek and silent Logitech keyboard for something with more prominent keys and (I’m sure) improved type-first-time accuracy.

I could probably save more time in my day than any AI could liberate just by investing £20 in a generic keyboard made for fat fingers, which I would happily modify by levering off the pointless Caps Lock key – the biggest waste of valuable real estate since Lex Luthor’s plan to split the San Andreas Fault in 1978’s Superman – to cut back on inadvertently shouting in emails and messages every so often.

However, like my haircut, such advice may be out of date.

We are entering the era of authenticity, where proving you are human becomes more valuable by the day, and which might even mean letting a few typos go just to show that I am in fact not a robot, since AIs tend to smooth out pesky typos.

At Glide, I am convinced we could remove nearly all text from our internal chat system and conduct all conversations via custom emojis, and still function just as well. It would be perfect timing at this point then, contemplating the idea of letting typos go, to use the Hide the Pain Harold emoji, probably one of our most used. Suffice to say I’m not for it.

Being human takes effort

My thoughts on a little imperfection being a good thing come after a double whammy of slow-clap AI hilarity welcomed me back to the real world this week regarding the trajectory of AI content on the web.

First up, I failed to stifle an ironic kyuk-kyuk at reading how scholars believe a nimble Chinese AI is running rings round OpenAI et al by producing superb responses out of a new model for a fraction of the cost of the bigger names, with a hint of suspicion that it might have shortcut the lengthy LLM learning process by harvesting ChatGPT’s answers amongst its efforts.

Surely OpenAI is up in arms about this? Well, it would if it could I bet. But is the maker of ChatGPT in any position to comment on the theft of other people’s work? So far all we have is CEO Sam Altman on Twitter cryptically going on about innovation being harder than copying. Quite right Sam!

I have no comment or insight on whether the new AI upstart did or did not simply point itself at ChatGPT to form its knowledge base. Nor on the OpenAI pot being unable to call the rival kettle black because to do so might prejudice its defence in separate legal matters.

What did spring out though was a reaffirmation of the issue of AI repeating work created by other AIs, whether it is innovative or correct, a battle arguably already lost even before Meta made it unwinnable by announcing its plan to rollout widespread AI-generated profiles on Facebook and Instagram – something which everyone knows has been an issue on social platforms for years anyway.

Perhaps to take on the surge in bot accounts, Meta will now create their own “genuine” fakes, complete with pictures, profiles, and the ability to write, respond, and share, and pump up engagement with the diminishing number of real people writing things there by cramming their feeds with interactions from faux-llowers. In due course, assumedly these fake accounts will be ingested into learning models and by that point we are only here to keep the lights on and the Meta share prices high.

It makes me feel that I am wasting my time correcting my errors and typos. Why bother, if they are soaring signs of authenticity? I already have to tapdance around the use of certain words like delve, captivate, and tapestry, for fear that they set alarm bells ringing that I am actually Robo-Rich, eater of RAM not lamb. Good words turned into the text equivalent of the Wilhelm Scream.

Then again, writing always meant watching your words. Avoid cliches like the plague, they said. "Frugal" was only ever written by motoring writers discussing fuel economy, and "sleek" was reserved for tech writers reviewing a new TV or fridge, and I avoided them like… ah.

One thing strikes me as certain: if the tech lords surmise that authentic human content is more valuable than any other, invest in whomever is selling the best services to de-AI AI-generated content. 

If not, I can recommend my technique of sitting at an awkward angle and the wrong height, with the keyboard about a foot too far away. Or getting a cat.

I hope you had a great holiday, and your 2025 is authentically human and happy.