Hacker News Reader

Comment by echelon | Hacker News Reader

Impossible. DNA has an estimated half life of only 521 years [1]. It's a fairly reactive molecule due to the number of reactions it has to undergo to fulfill its purpose.

I believe the T-Rex soft tissue [2,3] that has been discovered contained short polypeptide sequences. You can infer the codons from this information, but that elides promotor, signaling, non-coding, etc regions that will be of massive importance. Not to mention you won't get much data in the first place.

Hundreds of years from now, any dinosaurs we recreate will be the result of creative genetic engineering and imagination.

[1] https://www.nature.com/news/dna-has-a-521-year-half-life-1.1...

[2] https://m.phys.org/news/2019-11-mechanisms-soft-tissue-prote...

[3] https://www.smithsonianmag.com/science-nature/dinosaur-shock...

View on HN

> Hundreds of years from now, any dinosaurs we recreate will be the result of creative genetic engineering and imagination.

Not to mention machine learning. I wonder if we get a bunch of samples from different individuals, could an AI simply just piece it all together?

If the information is gone, it is gone.

I'm sure you could devise a "machine learning algorithm" that'll shove some random numbers in there, though.

It'd be like using machine learning to recreate unrecorded speeches of Marcus Aurelius or Alexander the Great word-for-word.

And in this climate I'm sure you could get some venture capital for that...

Given how many AI/ML companies infer meaning from meaningless data the information being gone isn't an issue. /s

Machine learning from what? Existing dinosaurs? The data is gone, what you're suggesting is that some type of AI is going to literally create life from nothing. Considering we aren't capable of autonomously driving a car down the road without killing someone, we've got a while to go before that happens.

> The data is gone

Correct

> Considering we aren't capable of autonomously driving a car down the road without killing someone, we've got a while to go before that happens.

These two problems are completely unrelated. Also, every form of transportation can result in injury or death, even walking and bike riding, so setting our goal at 0 deaths as a benchmark for success doesn't make sense.

This was more my commentary on attaching AI/ML to every single problem and expecting that to fix it, when we're VERY primitive in either technology.

... and the machine said "Let there be light!" and there was light. The machine saw the light and it was good.

Reply on news.ycombinator.com