Elodie Aishwarya Remoissenet

Founder of Barkley — behavioral intelligence layer for dogs. Writing on AI & what makes us human. 4 JRT, serial gamer, systems thinker — allergic to shallow AI.

[Elodie Aishwarya Remoissenet @getbarkley]

AI that can't survive drift, missing data, and a cold start isn't personalized. It's a flattering fiction, confidently wrong in production.
Surviving reality is the job.
New piece @HackerNoon 👇️
https://t.co/pmfnDZEqlI
#MachineLearning #AI #MLOps #DataScience #Personalization

[Elodie Aishwarya Remoissenet @getbarkley]

AI can copy your taste in seconds. It can never have stakes. And a choice that risks nothing means nothing. Closes the What Can't Be Generated series.

Full article 👇️
https://t.co/001wWsWmX2

#AI #Creativity #GenerativeAI #GenAI #VibeCoding #Ownership

[Elodie Aishwarya Remoissenet @getbarkley]

You can polish the surface forever. What a thing is built on always leaks through. What Can't Be Generated series.

Full article 👇️
https://t.co/M5OwGCKB57

#AI #SystemsThinking #Design #GenAI #VibeCoding

[Elodie Aishwarya Remoissenet @getbarkley]

The most dangerous AI output isn't the wrong answer. It's the wrong answer delivered with total confidence. What Can't Be Generated series.

Full article 👇️
🔗 https://t.co/KcDUXwz9Ss

#AI #GenerativeAI #CriticalThinking #GenAI #AiEthics

[Elodie Aishwarya Remoissenet @getbarkley]

You think you own your tools. Slowly, quietly, your tools start owning you. From the What Can't Be Generated series.

Full article 👇️
https://t.co/J2PSZNgT9j

#AI #GenerativeAI #FutureOfWork #Authorship #GenAI

[Elodie Aishwarya Remoissenet @getbarkley]

When everything can be generated, output is worthless. What's scarce now is everything that can't be generated: stakes, judgment, the real.

Finale of the AI × Human series.
Full article 👇️
🔗 https://t.co/YGAIiaC5GI

#AI #GenerativeAI #FutureOfWork

[Elodie Aishwarya Remoissenet @getbarkley]

Individual vs population baseline, head-to-head, synthetic data: ROC AUC 0.988 vs 0.935 · 100% vs 81% caught · ~34 days earlier. 30 seeds.

Reproduce → https://t.co/ovhbqxypbm
Release → https://t.co/jYsiib7CEW

#MachineLearning #MLOps #BuildInPublic #OpenSource #DataScience

[Elodie Aishwarya Remoissenet @getbarkley]

Behind the essay, the rigorous version: I formalized the Reference-Class Trap in a working paper, with a reproducible synthetic experiment — why population baselines fail the individual.

🔗 https://t.co/NjuYUL77Ni

#ResearchPaper #MachineLearning #MLOps

[Elodie Aishwarya Remoissenet @getbarkley]

AI won't become human by getting smarter. It'll become human the day it has a shadow — the part of us that isn't optimized, the part we'd rather hide.

Act III of the AI × Human series.
Full article 👇️
https://t.co/gD4CL4SYg8

#AI #Psychology #Philosophy #GenAI

[Elodie Aishwarya Remoissenet @getbarkley]

"Normal on average" can be dangerously abnormal for you.
Compare an individual to a population, and you erase the one thing that makes them an individual. The Reference-Class Trap — in monitoring, in health, in AI.

Full article 👇️
https://t.co/74ct9R3rFZ

#AI #DataScience

[Elodie Aishwarya Remoissenet @getbarkley]

https://t.co/jIYLgh4CcV

[Elodie Aishwarya Remoissenet @getbarkley]

https://t.co/saKjQTt4uy

[Elodie Aishwarya Remoissenet @getbarkley]

You spent years studying artificial intelligence.
You forgot the intelligence you already had.

Act II of the AI × Human series. Act I asked what AI is for; this one asks what humans actually are.
https://t.co/8BvySDKv6N

#AI #Humanity #Philosophy

[Elodie Aishwarya Remoissenet @getbarkley]

With #ClaudeFable5 back in the conversation, this perspective feels more relevant than ever.

Breakthroughs often don't come from solving known problems better, but from noticing dimensions we hadn't seen before.

In #AI, I'm increasingly convinced that many limits come not from computation itself, but from incomplete reference frames. We keep refining the model while sometimes overlooking the field it reasons within.

@AaronDinin Thanks for sharing this brilliant perspective.

Read the full essay 👇🏼
https://t.co/2dESAqNa0Q

#Startups #Entrepreneurship @claudeai

[Elodie Aishwarya Remoissenet @getbarkley]

Over four essays, I answered one question from four sides: what happens to us when the machine can do everything?

→ Act I: what AI is for.
→ Act II: what we are.
→ Act III: what AI lacks.
→ Act IV: AI governance.
To end, what stays valuable: https://t.co/YGAIiaC5GI

[Elodie Aishwarya Remoissenet @getbarkley]

The skill that matters is moving from technical to conceptual mastery — and that shift makes people uncomfortable.

I'm proudly AI-assisted. And I'm done apologizing for it.
https://t.co/ZHUQgRGZEd
AI×Human series. (#19 on @hackernoon TechBeat.)
#AI #GenerativeAI #FutureOfWork

[Elodie Aishwarya Remoissenet @getbarkley]

Most AI products skip the data architecture and jump straight to the interface. That's exactly why they fail.

Selected as a Top Story by @hackernoon
https://t.co/DX7e7a7tm7
#AI #LLM #DataArchitecture #DataScience

[Elodie Aishwarya Remoissenet @getbarkley]

Compared to the population, you look normal.
Compared to your own history, you have changed.

Not a bias problem. A reference-class problem: the wrong unit of comparison.
Normal for the group, abnormal for the self.

via @DDInvestorHQ

https://t.co/9owX12UJ71
#AI #AIEpistemics

[Elodie Aishwarya Remoissenet @getbarkley]

Talent acquisition loves the vocabulary of the misfits. It just doesn't want the misfits.
Same story. Same skills. Same person. Only the signal changed.

The essay that started the series ↓
https://t.co/Mg2n8eJDer

[Elodie Aishwarya Remoissenet @getbarkley]

You can't de-risk a breakthrough. You can only de-risk your way into a purgatory of endless sequels.

We took a discipline built to protect what exists — and aimed it at creating what doesn't. In creation, risk isn't a bug. It's the substance.

The De-Risking of Everything ↓