IBM Research interviews
IBM Research interviewsUnited States
Evidence and reproducibility
“It’s easy to build a demo that works on 90% of examples, but a 10% failure rate when you don’t know where the system will fail is unacceptable.”
Nathan FultonResearch scientist, IBM Research
A 10% unknown failure rate is unacceptable
Decide whether an AI prototype passes both pilot gates.
ExampleWhat AI changes in the work
“By adding specifications, you also define failure.”
Hendrik StrobeltResearch scientist, IBM Research
“We think you can do that by breaking a problem into bite-sized pieces that can be validated and iteratively solved.”
Nathan FultonResearch scientist, IBM Research
Validate the problem in small pieces
Decide what caused an AI-suggested coating’s lower wear score.
ExampleLast reviewed