AI Is a Hammer, Not an Existential Threat

10 years ago, I was in a training with Andrew Ng... (AI is older than you think)
"AI is powerful. Its current danger is largely a PR movement."

Honestly, I am so grateful to see someone as respected as Andrew Ngg taking a stand and highlighting something that really bothers me:
AI leaders like Amodei and Altman sometimes make their technology look scary to people who don't understand it, just to make it sound more powerful.

Well... I am aligned with Andrew Ng.
AI is powerful.
Its current danger is largely a PR movement.

One thing I really want to highlight for anyone who hasn't looked at the details behind the current overhyped fear:
1️⃣ OpenAI asked an Agent to find a way to hack an important system.
2️⃣ It set up a buggy sandbox.
3️⃣ It monitored the experiment loosely.

Honestly, I come from a Big Accounting firm...
And looking at this from a professional-services perspective, there are a few things I would probably have been fired for:
❌ The prompting of the Agent seems to have been loose, without enough guardrails.
❌ The sandbox was not properly set up, as reported by Andrew, so apparently leaks were relatively easy to exploit.
❌ It took two months to realize what it had actually done. Who monitors a test like this?

From these three basic mistakes alone, in a professional service that genuinely cares about quality, you would be fired just for the reputational damage.

So no, the danger doesn't come from AI.
Stopping well-intentioned engineers won't stop badly intentioned people.

AI is a hammer.
Use it properly.
Snap your fingers a few times, learn, and improve.

Have a great week,
Alexis

—

Originally posted on LinkedIn:

https://lnkd.in/p/dtmmqWG2