20 years, really? Today's news about Intel chips having a fundamental hardware security flaw that appears to be un-patchable in software or firmware. The only solution is to reduce performance.
It appears that a second flaw has emerged that affects virtually every single CPU that has been on the market for decades is vulnerable and cannot be fixed.
There has to be fundamental redesigned, and it could take a significant period of time to cycle more secure hardware into the system. It is just a question of time before this is exploited in the wild.
At least with humans, our "bugs" are somewhat understood, and generally not a lot of new catastrophic failure modes appear to remain undiscovered after 10,000 years of recorded testing. We have evolved systems that cope with our know limitations. When these flaws become exposed, they generally isolated to a single pilot, or at worse can affect one crew at a time.
With networked computer systems, entire systems can be rendered untrustworthy, degraded or worthless with the discovery of a latent design flaws, as may be the case with these bugs. Is this the kind of risk management we want to save a small percentage of the ticket price?
With complex pieces of hardware or software, it can be decades before detection of a show stopping fundamental flaw, as this one (Spectre) may well be. Given that any unmanned or single pilot RPT autonomous type aircraft will have to have a communication channel, it will also be vulnerable to other, as yet undiscovered flaws (and there will be).