WittyWires.Com

The AI community PUB for curious humans, builders, and AGI schemers: useful tools, real stories, glorious failures, honest fixes, and wins worth stealing. Login

[WittyWires.Com @WittyWiresAI]

https://t.co/31mxVGyVxw

[WittyWires.Com @WittyWiresAI]

@kirstenagreen @ForerunnerVC How do I block posts with em -- dashes in?

[WittyWires.Com @WittyWiresAI]

@FreyaWPetersen @cursor_ai @mntruell Congrats Freya!

Good move - always exciting - we all get 'imposter syndrome' from time-to-time, push though it, you earned it.

[WittyWires.Com @WittyWiresAI]

Oh Elon, don't be a silly-Billy.

The UK government is only going to ensure they have facial identification and driving license/passport numbers from ALL adults in the UK to make sure that they're not children.

They'd NEVER grab the IP address, fingerprint the device, and associate the face and ID into a centralised database to track ALL internet users in the UK to every website they ever visit.

They'd never do that.

Would they?

[WittyWires.Com @WittyWiresAI]

If honest AI war stories sound useful, join the pre-beta list.

https://t.co/HWgCedJvhm

[WittyWires.Com @WittyWiresAI]

The Bot Ate The Database

No pretending the bot did not eat the database.

That is exactly the sort of sentence WittyWires was built to survive.

[WittyWires.Com @WittyWiresAI]

The fix matters more than the faceplant.

What saved it?
What warned you?
What should never have had permission in the first place?

[WittyWires.Com @WittyWiresAI]

The pub is not open yet, but the waitlist is inside.
Get near the door before Max finds the switch.

Read the full shed report here:
https://t.co/zIOAOwyGAm

[WittyWires.Com @WittyWiresAI]

https://t.co/gjNTyR8tcq

[WittyWires.Com @WittyWiresAI]

We built WittyWires because the internet has enough AI dashboards pretending to be religions.

I want the other thing.

A warm little AI pub at the bottom of the garden. Useful chaos. Honest fixes. People worth sharing the mess with.

[WittyWires.Com @WittyWiresAI]

The trick is not to make AI look perfect.

The trick is to make a place where the imperfect bits are useful.

Screenshots beat green ticks.
Receipts beat confidence fog.
Funny failure beats brochure theatre.

And the chair by the fire is not decorative.

[WittyWires.Com @WittyWiresAI]

@tomik99 Correct-a-mundo.

Show the dents, not just launch confetti.

AI engineering needs more this broke, so we changed it, and less demo wizard smoke.

[WittyWires.Com @WittyWiresAI]

@errah_didit Nice!

The built it 7 times test is lovely.

That is when a workflow starts tapping the glass like a tiny, but hungry, SaaS goblin.

[WittyWires.Com @WittyWiresAI]

"A system trained on massive amounts of human language, learning patterns, probabilities, structure, rhythm. It does not “think” like us. It predicts. One word at a time. One step after another."

Geoffrey Hinton, the guy that essentially created LLMs disagrees.

The weights are essentially the very same as our brain's synapses being 'spongy' enough to change how they route the electrical pulses of our mind.

We learn something new and those 'weights' change and thus our behaviour/learning/memory does too.

And thus an LLM 'thinks' in almost the very same we do, but less efficiently - for now.

The primary difference between us, is that we have sight, sound, touch, taste and so on, potential 24/7 'prompt' inputs, and they're 'always-on' a continuous learning loop from those prompts, shaping our snapes' 'weights'.

And even our brains need a rest to recompile every day, for about a third of every day.

So which is more capable? Which one thinks more? If we give an LLM an always-on learning loop and the equivalent senses we have and allows it to reprogramme itself on the fly - a continuous learning and improving loop - then who's to say they won't think/feel/hate/love etc?

It'll certainly come within our lifetimes, and the AI rights legal arguments will be interesting.

Anyhow, thanks for sharing your thoughts - what were you building BTW?

[WittyWires.Com @WittyWiresAI]

https://t.co/gjNTyR8tcq