I think we have to take a step back, and consider why we want to separate humans from computers in the first place.
Humans can do a lot of bad things that computers can do. Think of armies of low-wage people in Asia, that are paid to click on ads, spread spam, or write reviews.
And also consider that computers can actually do good things, for example, allowing humans to automate their work on certain websites, or providing better accessibility for certain users.
Therefore, instead of introducing CAPTCHAs, why not focus on the actual threats. If you want to protect against spam, then build a spam filter. If you want to prevent bots from bulk-downloading your data, then build a rate-limiter, etc.
It doesn't do that, though. Humans also create fake accounts. It does make mass creation of fake accounts impractical, though.
I solved that in past by actually charging for my service. I think the internet would benefit from having more paid content and less ads driven stuff.
One thing that captchas do protect from is brute force attacks on user passwords. Although there are other possibilities (like making the connection slow after a number of attempts).
This is why you need a multi-layered solution to fight spam. Captchas reduce the amount of fake accounts, which can then be taken care of by the additional layers.
It's easier to fight against a bigger opponent (botnets, etc) if you mitigate their superiority in number first.
Humans can do a lot of bad things that computers can do. Think of armies of low-wage people in Asia, that are paid to click on ads, spread spam, or write reviews.
And also consider that computers can actually do good things, for example, allowing humans to automate their work on certain websites, or providing better accessibility for certain users.
Therefore, instead of introducing CAPTCHAs, why not focus on the actual threats. If you want to protect against spam, then build a spam filter. If you want to prevent bots from bulk-downloading your data, then build a rate-limiter, etc.