André Thénot

So… first 80%: zero effort, last 20%, 80% effort. That's a 20% savings right?
Not so fast because picking up where AI left off may make those last 20% take MORE than 100% of the effort. eg. cleaning up an ugly codebase, or an artist trying to recover an AI sketch…
(2/2)

11.07.2025 01:57 — 👍 0    🔁 0    💬 0    📌 0

AI feels magical in that it makes easy tasks that were completely out of reach because their skill isn't in our wheelhouse.
On the flip side, that breeds disdain for the real experts because we get 80% of the result with zero effort.
As the saying goes, the last 20% takes the remaining 80%
(1/2)

11.07.2025 01:53 — 👍 0    🔁 0    💬 1    📌 0

Adversarial AI: using it like a Devil's Advocate to poke holes in my own logic. Since I liken AI to the "wisdom of the crowds", this mode of usage helps catch when I miss something that's totally obvious to everyone else.

09.07.2025 16:40 — 👍 1    🔁 0    💬 0    📌 0

I'm still discovering new areas where I can trust AI-generated code; getting a good amount of success when I give a very specific task with little room for interpretation.

27.06.2025 13:35 — 👍 0    🔁 0    💬 0    📌 0

AI already has some really practical benefits in areas that are massive pattern matching exercises.

At the same time, "AI-free" will become a selling point in certain areas. This has already occurred in commissioned art work but will surely extend to other domains.

10.06.2025 14:57 — 👍 0    🔁 0    💬 0    📌 0

The cockpit of an Airbus A350. Tactical view to fly an aircraft following a well understood process.

Jira Kanban board. High-level story telling but obscuring a lot of nuance that affects decision-making.

When building products that tell a story, interface simplicity is key, to not overwhelm the user.

But power-tools benefit from high information density, to display more nuance onscreen. That begs the question: what's the signal to noise ratio on our day-to-day display? Usually poor. 😢

10.06.2025 12:01 — 👍 0    🔁 0    💬 0    📌 0

I'm trying to visualize the journey from LLM to agents and beyond. Does this match people's journey of learning ? Or is it more the industry talk of people experimenting ? Anyhow, would love to hear your feedback ! #aiengineer

06.02.2025 10:35 — 👍 9    🔁 2    💬 2    📌 0

As we get a rational understanding of the areas where AI is good at automating—and where its failure modes are above the risk threshold—we'll converge to a new equilibrium of humans and machines each respectively doing what they are good at.
(3/3)

10.06.2025 00:33 — 👍 0    🔁 0    💬 0    📌 0

Less predictability because we're still trying to get a good grasp on failure modes. That's always the interesting aspect in any system. But we're making progress.
(2/3)

10.06.2025 00:28 — 👍 0    🔁 0    💬 1    📌 0

Make humans do what humans are good at, and make machines do what machines are good at.

Machines are good at automation and systematizing. AI blurs the lines because it can automate a whole different class of problems, and more complexity = less predictability.
(1/3)

10.06.2025 00:27 — 👍 0    🔁 0    💬 1    📌 0

Hello Word! First post here even though I had this account for over a year. I don't expect this account to be super interesting, just writing about things I'm playing with—currently AI.

09.06.2025 17:07 — 👍 2    🔁 0    💬 0    📌 0