Any change analyzed in a vacuum is risky. The trick is to compare it to the alternative, and see if it is more risky. The alternative to GMO is applying random mutations, and then selecting a strain that has the behavior we want. This has much higher risk than changing only what we need, since many other components could be changed at the same time too, and each of them also have these many interactions.

Honestly, Taleb should stick with what he knows: finance. In other fields I trust his opinion about as much as I would any other Wall Street banker.