Igor Khrol: Тестирование искусственного интеллекта: с какой стороны подступиться

Testing community is discussing how machine learning can find bugs and replace testers. Meanwhile, the reality is more boring and not so futuristic. Engineers are releasing machine learning and artificial intelligence services that should be tested. Based on real experience of delivering such products I want to share how different such testing is from regular solutions and what are possible problems. We’ll go through: - metrics used to measure the quality of the solution - what specifics bugs can be observed there - and set of tips and tricks that will help you to save time and money during ML systems development Bio: I’m a software engineer with a strong math background. More than 13 years of industry experience in different scale companies on various roles: engineer, team/tech lead, manager, consultant, trainer.