How were later robots able to emotionally harm humans?

Question

In the short story Liar!, part of I, Robot, a crucial plot point is that Herbie, the mind-reading robot, was bound by First Law not to harm humans, and being that he understood emotions, this included emotional harm. This manifested itself (in part) in telling the person exactly what they wanted to hear.

Later robots did not face this problem.

Assuming Stephen Byerley (from Evidence) was actually a robot, he certainly caused emotional harm to a lot of people, simply by running for office.

In The Caves of Steel, R. Daneel Olivaw once even causes Elijah Baley to pass out, after Lije's accusation that Daneel was actually human was proven false. Throughout the series, Daneel has no qualms about telling a person they're wrong.

R. Daneel and Byerley, being robots designed to pretend to be human, are certainly aware of human feelings. How then did they not face the same problem as Herbie?

Answer 1

You seem to be forgetting something- Herbie wasn't just capable of understanding human emotion, he was capable of reading their minds.

Even more importantly, he wasn't programmed for that. Robots capable of reading emotions that were programmed for it could, of course, deal with that by prioritising properly. They have the programming to handle it, else they wouldn't have even gotten to alpha testing. Herbie wasn't programmed to deal with being able to read minds.

I think if you suddenly gained that ability you, too, might have trouble dealing with you saw. Herbie's understanding of emotions was somewhat primitive, but he was compelled by his existing programming to act. He was able to stop people from feeling bad by telling them things (even things that weren't true). This worked in the short term, but hurt them even more in the long term. He couldn't deal with that.

Answer 2

Herbie was a very early design robot - US Robotics and Susan Calvin started in the 21st century [timeline].

The stories involving R. Daneel Olivaw and Elijah Baley occur 1500 years later. Daneel is a very advanced design robot (who eventually co-develops the zeroth law). Daneel has also been in operation for sometime before he is introduced in the books, whereas Herbie was a brand new design fresh from the production line.

Once you couple more advanced positronic brain design with better training and education, it would simply appear that Daneel (and other robots later than Herbie) are better at distinguishing between relative harms.

Telling someone that they are wrong has an immediate harm associated with it - but allowing someone to labour under a false belief could be considered the greater harm.

I don't really have much of an explanation for Byerley in Evidence or The Evitable Conflict - except that it is an open question of whether Byerley is a three-laws robot or not.