A series of articles on AI Ethics.

Starting today, I’d be covering three major articles on the ethics of artificial intelligence. I link them from HBR and raise very important issues. As usual, I’d highlight the key principles in discussion and weave in my ideas around them. Some of them are pay walled, and while the bias in inherent (because the industry sponsors them), the arguments cannot be dismissed outright. The industry linkage is troublesome because of specific agendas inherent in the way they weave subtle marketing pressures in the context. Nevertheless, in absence of a coherent research in the ethical AI (beyond the obvious disparities in people of color and attendant brouhaha), it is also essential to look at these articles from the eastern perspective.

Multinational corporations span the globe with significant presence in several markets and hence a leverage to push the governments towards their specific agendas. While China pushes out their own gimmicky version of robots in some science congress to showcase its “prowess”, these solutions do little to either advance the cause of science or have a significant hype towards vanity metrics. Likewise, the promise of “personalised medicine” intertwined with the vast pools of genetic data has failed to make a dent in what physicians are concerned with-overall survival. I have written about these previously- how the university hospitals have started embracing trends to push for their “research-oriented” treatments in specific geographies. They overstate the promise of “personalised” goals whereas the outcomes remain uncertain.

Does it represent the lost cause? Does it mean that investments are useless? Does it mean that enterprises are not geared towards the adoption?

No. None of it. AI and machine learning work best on structured data, and I have been pushing for semantic web. I agree that there are significant trade-offs and the design principles (datasets) don’t agree with the working solutions. The enterprises (and healthcare organisations) have also failed to devolve their ideas around what data science can actually do for them. They cannot organise their business proposals/workflows around central principles of improving efficiency and automation. Often, there is no principal “returns on investment” plans (ROI) for the shiny new hires in their data science departments. The marketing department, though, pushes out embedded stories in rag publications that dot silicon valley- usually journalists for hire. It is a sad reflection of the state of affairs as they often “dumb down” the complex underpinnings of enterprise workflows.

As the ideas remain hazy around the validity of the software constructs and its end goals, there is an unnecessary fear psychosis around AI coming for the jobs. Large-scale studies around automation and job losses will eventually miss out several sub-contexts and cannot generate nuanced arguments about what exactly is under the chopping block. It leads to unnecessary speculation and therefore the rise of AI ethicists.

Some authors and “researchers” have made their careers out of the idea of “ethical AI” and how the machine learning algorithms are biased against specific communities or neighbourhoods. They took these arguments to a fever pitch in some countries and remain the staple of tweets that often go viral. Ethics is an abstract science that leads itself to myriad interpretations. Therefore, no one answer can subsume the “controversies”.

I believe that the affirmative actions and the crony capitalism have made entrenched lobbies that prevents a sensible debate. This opinion may not go down well with some of the readers but it has to be seen rationally and without getting triggered about “people of colour”. The historical unstructured data will yield only flawed assumptions. Therefore, it is a clarion call for a better research in defining accurate datasets, open source algorithms, and involved populace on how the assumptions have been made in the final output. It calls for a clear-headed policy and political debate around the importance of artificial intelligence.

I do hope that you’d find the following articles interesting and hopefully would help us to understand and adapt them in hospitals.