Go to any conference or event these days and the number one buzzword is AI. However, despite all the excitement, AI is still talked about in fairly vague terms – with few people able to pin down exactly what they mean by AI. New research from some of my Elsevier colleagues examined this phenomenon and found a significant gap in understanding across several industries when talking about AI.
Our research reviewed all the words and phrases commonly used by different groups to refer to AI (industry, teaching, media, and academic research); it identified 1,033 AI-related keywords being regularly used across the various fields. Yet, only a miniscule six of these words were commonly used by all four groups – so while we’re all talking about AI, we don’t have a common language with which to communicate.
Universal problems require collaborative solutions
This is a problem because making AI work optimally for all requires collaboration across sectors. Take the ethics of AI as one example. Algorithms are becoming a much greater part of our daily lives – which comes with a host of ethical issues that corporations must address. When corporate researchers speak about AI, they are usually speaking about algorithms to improve efficiency or output, rather than ethical considerations. So those building the algorithms need to talk to those with the scientific know-how, to design effective AI models that are free from bias.
As Elsevier’s Dr Jabe Wilson has highlighted in the recent Laboratory Informatics Guide 2019, which you can access here, there is a challenge around how to ensure AI algorithms we use to predict the efficacy and safety of new drugs aren’t biased due to incomplete, inaccurate data, or inaccessible data? Dr Wilson notes, “‘What we are doing at Elsevier – and many people are struggling with this – is making sure that the semantic data and semantic indexing can throw its hands around all that complex data and put it in a place where people can access it.”
Since academic researchers have been far more active in these areas that corporate researchers, there is clearly scope for more public-private partnerships to tackle these sorts of issues. The more that we see interaction between business, government and academia, the more progress we’re likely to see in developing AI systems which are both effective and responsible.
In order to realize this improved level of collaboration, all stakeholders involved with AI development should aim to adopt universal standards to improve communication. To help this cause, we’ve donated our Unified Data Model (UDM) to the Pistoia Alliance (where our very own Tim Hoctor sits on the board) in order to help all pharma companies manage their data in the same way – we hope to see other industries and organisations follow suit.
For further information or a discussion on how Elsevier’s decades of experience in data management can help meet your AI needs, contact our sales team today.
Nicki Catchpole
As a professional with over 14 years of experience in strategy development and partnership management across a variety of industries, Nicki’s latest role as a Senior Manager, Segment Marketing at Elsevier applies her skills to the area of drug discovery and development in the Pharma and Biotech industry.
In this capacity she is focused on understanding biopharmaceutical R&D challenges and turning them into opportunity to further Elsevier’s ability to serve industry executives and the professionals who innovate in the drug discovery and development space.
Nicki resides in New York City and holds a BA in English Literature and Mandarin Chinese from Carleton College in Northfield, MN.
Source link
