automated decision-making systems

what is a process or decision that used to be made by humans and is now automated by sophisticated algorithms?

apart from obvious real-world examples of automated decision-making systems (adms) of self-driving cars and trading, where 85% of all transactions in the foreign exchange markets is conducted by algorithms alone (demetis, 2021), bidding bots for freelance project posts represent another example of adms produced for a niche market, in which i was participating during my high-school days. in essence, these bots are refreshing a dashboard of project posts and automatically making bids on a specific project that fulfils the custom criteria defined by a freelancer, who built or purchased this kind of software solution.

could there be unintended consequences of using algorithms? who could be harmed and how?

consequences of using adms range from less probable with a low impact to more probable with a high impact on business processes, society, and the environment. the media has been consistently showing many examples demonstrating that self-driving cars are dangerous for human lives. the vast majority of these situations are caused by problematic and unexplainable decisions made by ai-driven computer software.

on the other hand, mass and extensive use of bidding bots can undoubtedly result in an unregulated monopoly or freelance projects being assigned to incompetent people, diminishing the purpose of utilising freelance platforms.

to conclude, we have to answer one important question. should we focus on creating static and transparent algorithms, e.g. expert systems, which can explain their reasoning but may not solve the given problem due to model simplicity, or dynamic and black-boxed models using machine learning, e.g. deep neural networks for reinforcement learning, which can solve the given problem but are very unlikely to produce any explanations of their actions? (marabelli, newell, and handunge, 2021).

references

demetis, d., 2021. algorithms have already taken over human decision making. [online] the conversation. available at: https://theconversation.com/algorithms-have-already-taken-over-human-decision-making-111436 [accessed 14 september 2021].

marabelli, m., newell, s., and handunge, v., 2021. the lifecycle of algorithmic decision-making systems: organizational choices and ethical challenges. the journal of strategic information systems, 30(3), p.101683.

important note: this article is was used for the second assignment for using data-science responsibly course at london business school