AI generates code faster than we can review it.
A USENIX study found 20% of AI-suggested packages don't existโand 43% of those fake names are reproducible enough to weaponize.
It's called slopsquatting.
https://t.co/t5FaJmo5Zl
@steipete@DustinDavis True I have a rust version for highly specialized use cases. Lessons learnt is the bottleneck is the inference unless u point to models on the chip. In that case it's blazing.
In today's market, slow and steady loses the race. Build rapidly, iterate relentlessly. First-mover advantage isn't about perfection, it's about speed. Your moat is being a couple of steps ahead of everyone. #ProductDevelopment#Agile#AI#RapidPrototyping
๐ค Want to be more efficient? Join us on Wednesday, April 5 for "GitHub in my day job". Our experts will show you how GitHub can help you streamline your development cycles and enhance your overall productivity. ๐ก https://t.co/I0G9EPKiaK
Did you know most ML algos have a built in bias that reflected from it's training data?
We need to acknowledge this inherent bias and build responsbile and fair systems. https://t.co/XjibP9tisW