The Challenges of Artificial Intelligence

Data and technology were created by humans to reflect and advance human values. It is important to not trust blindly the outputs of automated systems without vetting the accuracy of the inputted data and the decision-making process that is taken.

“Bad data and methodological mistakes can easily lead to erroneous results"

Even when the characteristics such as race, religion, gender and ethnicity are fully eliminated from models, there could be a chance that other features that are correlated with this kind of characteristics may be included and introduce the same bias. Researchers can be blind to issues with this characteristics that aren’t easily captured by numbers.

The Effects of Discrimination

While price discrimination based on race, religion, nationality or gender is illegal in the united states, enforcing these laws is a lot more challenging when it comes to e-commerce, where the pricing is defined by the algorithm.

Malicious AI

AS machine intelligence becomes more powerful it increases the risk of attacks that compromise security infrastructure which protect our resources and communities. AI will be used to multiply the effect of malicious campaign and to broaden the types and number of possible targets.

AI may alter the arena of cyber-attacks: increasing the efficacy and precision.

The use of technology can have a big social impact.

Source: Yao, M., Jia, M. and Zhou, A., 2018.Applied Artificial Intelligence. pp.609-673.