Survey Predicts Three-Year Return On AI In Drug Discovery

2653
AI should be able to help in drug discovery, said AI Trends readers in response to a survey.

By Allison Proffitt, Editorial Director, AI Trends

Nearly everyone believes that AI can help in drug discovery within the next three years, but the devil is in the details.

That’s according to the new survey on AI in Drug Discovery from AI Trends. We asked readers about how they are using AI in their jobs and what opportunities and challenges they see ahead for the technology as applied to drug discovery. Almost unanimously respondents believe that AI can assist in drug development.

But the specifics get murkier. This year about one-third of our survey-takers represented biotech and pharma and one-third hailed from academia. The rest came from hospitals, government labs, technology providers and more. 59% reported that they use AI already in their role, with machine learning, pattern recognition, deep learning, and image recognition leading the applications.

Compared to our 2018 survey, personal concerns spread more evenly across many options. We asked if survey takers had AI-related technology concerns related to their own roles. About a third of respondents expressed concern about data security, data quality, and fear that the technology would make errors. But almost equally concerning was overwhelm from features that users either didn’t understand (30%) or didn’t need (23%). Time and cost were still cited as a concern, as was job loss. But 14% of respondents—down from 20%–expressed no concerns at all.

Early discovery, new target discovery, and data mining led the list of areas in which the respondents expect significant contributions in the next three years—each with over 60% of respondents choosing them. But no research area received less than 23% of the vote, indicating a hopeful outlook for short term gains from the technology.

When asked to pinpoint hurdles to AI-enabled drug discovery, trustworthiness led the field, receiving nods from 53% of respondents. Regulatory and cost hurdles each accounted for about 45%. But the next most frequently chosen option was “Biological Variability,” highlighting the challenges inherent in drug discovery for AI or any other enabling technology.

See the full survey results including 2018 findings.