Generally the change is n jobs paying $w / hour doing X being replaced by pn jobs paying $rw / hour doing Y. The markets generally ensure that p*r < 1 long term (for constant market size); however, the values of p and r, and the economies of scale of new technologies have a profound impact on their overall impact on important societal outcomes like inequality.
For example, consider a technological advance which has relatively low barriers to entry, such as voice over IP. In this case, r < 1 (people doing VoIP work get paid less, on average, than people at telcos before) but p > 1 (access to telecommunications technology is easier, so more people do it, even at fixed same market size). This creates more jobs and reduces income inequality.
However, in other cases, new technologies have higher barriers to entry than the existing manual option, p < 1, r < 1; medical equipment that is prohibitively expensive to manufacture and operate except for big players; small clinics close down and a few mega-clinics open up; while production is more efficient overall, the market is less efficient, and profits are higher for the few that remain in business. There are less jobs and more inequality.
The best possible outcome is when technology that makes jobs obsolete is low-cost and accessible to everyone; jobs are lost, but people don't need to work as much to make ends meet.
The worst is where new technology creates economies of scale that mean few big companies control major production but don't need many workers, but still charge only marginally less than the more labour intensive alternative; jobs are lost but prices don't go down by much. The only solution to maintain income equality is government taxation of those big companies combined with government public good expenditure (e.g. hiring the people who lost their jobs to do research).