A Sense of Doubt blog post #3391 - AI Chindogu - via ELLIS

A Sense of Doubt blog post #3391 - AI Chindogu - via ELLIS


Some thoughts on AI.

More Warren Ellis.

Thanks for tuning in.

Weather Systems Orbital Operations for 19 May 2024

ORBITAL

THE MANDATED AI HOT TAKE

I’ve avoided writing about AI too much. I hate the term "AI",” for a start. It’s not intelligent and it’s frankly not always artificial. Remember Amazon’s “AI-powered” cashier-less shops? The majority of those sales were managed not by AI, but by humans in cubicle farms in India. LLMs rely on human curation (as well as human art, thought and breath).

AI cannot simulate futures. If a machine can’t think ahead, it’s not an intelligent machine, just really good at running spreadsheets. To mangle a recent metaphor: it doesn’t know if it should open the pod bay doors or not, because it's not capable of simulating the multiple possible outcomes of letting Dave back in or not. It cannot wonder about things.

Cats will wonder about things. As ever, you need to be worrying about the people, not the robots.

And it’s all appallingly expensive and runs at a loss.

Anyway. We are at the point where I am finding some interest. I’ve been testing Perplexity AI for search, because I need easy access to knowledge engines for my work, and it ranges from useful to bloody hell that’s pretty good. But the thing that’s caught my eye this year? The chindogu era.

Chindogu is a Japanese term: the weird and “un-useless” invention. An unusual and ingeniously devised tool that does one fringe operation, poorly. It’s not useless? But nearly so. Which seems to be the general opinion about the Humane AI Pin, and, to some extent, about the Rabbit R1.

One classic chindogu the solar-powered flashlight with no storage battery.

Right now, even Google are shipping software that might one day do the things they’re talking about, but doesn’t right now and won’t in the immediate future. The Rabbit R1 appears to have shipped without core functionality in place, which is a shame. The Humane Pin is, for some people, a terrific idea on its face, but so is the chindogu Baby Mop, a duster attached to a baby so it can clean the floor while it learns to crawl.

(Hey, I come from a people who used to stick children up chimneys to clean them, and frequently threatened to make my child do the same.)

I suspect there may be a bunch more of these. Android phones will try to eat the standalone AI device’s lunch before it’s even ordered (the Rabbit R1 was supposed to be able to order lunch but it can’t), and Apple would like to but they have to weigh improved AI access against their traditional concerns about security, so I suspect Siri will remain that thing we only use to set timers or reminders sums with. But the Humane and the Rabbit have probably opened the doors, especially if AI compute gets cheaper, and there could well be a brief era of AI chindogu, coming to an Amazon storefront near you via container ships from China. That’s amusing stuff for me.

And, as I said, AI-enabled search is getting really interesting now. I half-caught a news headline the other day, and asked Perplexity for a brief on the current political situation in Slovakia because I didn’t have enough hands free to type. I got a full abstract that included the shooting incident that happened a couple hours previously. And links to six sources for further reading. That was impressive.

Anyway. The intersection of AI with search and chindogu is a curious space to explore.

LOW POWER MODE: I sometimes put the blog in what I call LOW POWER MODE. If you see this note, the blog is operating like a sleeping computer, maintaining static memory, but making no new computations. If I am in low power mode, it's because I do not have time to do much that's inventive, original, or even substantive on the blog. This means I am posting straight shares, limited content posts, reprints, often something qualifying for the THAT ONE THING category and other easy to make posts to keep me daily. That's the deal. Thanks for reading.

+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

- Bloggery committed by chris tower - 2405.31 - 10:10

- Days ago = 3255 days ago

- New note - On 1807.06, I ceased daily transmission of my Hey Mom feature after three years of daily conversations. I plan to continue Hey Mom posts at least twice per week but will continue to post the days since ("Days Ago") count on my blog each day. The blog entry numbering in the title has changed to reflect total Sense of Doubt posts since I began the blog on 0705.04, which include Hey Mom posts, Daily Bowie posts, and Sense of Doubt posts. Hey Mom posts will still be numbered sequentially. New Hey Mom posts will use the same format as all the other Hey Mom posts; all other posts will feature this format seen here.