As a hobbyist honeypot admin I see some of the day to day hack attempts, script kiddies scanning ranges, researchers scanning/monitoring and the sophisticated black hats we see today.
The internet of things needs serious consideration before deployment, these days too many things are rushed out without testing mainly due to management or deadline goals.
In the last few years or so we have seen proof of exploitation’s/concepts that are dangerous not only to computers but human life (stuxnet, Bnarby Jack (RIP) just before he was about to disclose his research to BlackHat2013 about pacemaker wireless exploit and died just before)
Yet we are looking at connecting fridges to the internet so they can order milk as it runs out, lol just look at the possible DoS/DDoS effect imagine gretting the delivery driver with 20,000 pints of milk!
At the moment computers are controlled by humans who are prone to errors, the time will come that computers arent but the code is legacy (especially in embedded systems!, shellshock!)
Exploits always exist so are we really ready for Self driving cars? Especially as we already see security flaws, BMW have just patched a design flaw see http://grahamcluley.com/2015/02/bmw-security-patch/.