In a world of Artificial Intelligence
In a world where companies like Netflix, Spotify, and Klarna set the standards for our expectations as customers, it is safe to say that artificial intelligence (AI) and machine learning will grow stronger. AI solutions use to calculate the likelihood that you will appreciate certain songs, which scenes to include in movies to give them high ratings and pinpoint what a fraudulent transaction looks like to provide smoother payment solutions. IBM started exploring the area of AI many years ago, in the 90s developing Deep Blue, the first computer which won over a human in the board game chess, and in 2011 having developed Watson, the first computer system to win Jeopardy while competing against ruling Jeopardy champions.
In later years the importance of working actively with the ethical aspects of AI has become more evident. IBM puts a lot of money and resources into generic research, in 28 years in a row, the company has filed for more patents than any other company in the US. Significant areas for IBM research are AI and Ethical AI (IBM, 2019). Ethical AI is an important research area as we have seen many examples of where AI systems are not automatically performing ethically. Unfortunately, there are many examples of AI systems acting discriminating. We have read about recruitment systems being favorable to men (Gibbs, 2015, Dastin, 2018), healthcare systems that discriminating black people (Ledford, 2019), white offenders being more likely to get shorter sentences than other offenders (Thadaney Israni, 2017), and women being granted lower loans or credits than their male peers (Telford, 2019).
The year 2020 became essential for the discussion around the use of facial recognition. Both Russia (Reuters, ND) and China (Jakhar, 2020), were accused of using such systems unethically in their surveillance of people. In 2020, IBM announced the decision to stop develop general-purpose facial recognition systems to prevent the use of facial recognition systems for mass surveillance, racial profiling, or any other type of violation of basic human rights and freedom (BBC News, 2020).
So, how can research help prevent unethical AI?
IBM works with five focus areas (IBM, ND). This is to consider ways to make AI systems’ reasoning transparent and understandable to humans (Explainability) to reduce and discover bias in data which lead to discrimination (Fairness) and to prevent security issues such as hacker attacks (Robustness). Furthermore, to make the shortcomings and the strengths of AI systems visible (Transparency) as well as to find ways to ensure your private data stays private even as it is being used by AI systems (Privacy). IBM believes in a future of AI given that morals and ethics lay the foundation for how these systems are developed and used further. If you are interested in learning more about ethical AI, take a free course at IBM’s no-cost learning platform Open P-TECH.
Everyday Ethics for Artificial Intelligence