Can AI ac­cel­er­ate drug R&D? J&J of­fers up some mol­e­cules to try it on

Lon­don-based Benev­o­len­tAI be­lieves it has built the kind of ar­ti­fi­cial in­tel­li­gence tech that will al­low it to iden­ti­fy and de­vel­op drugs faster and bet­ter than any group of mere sci­en­tif­ic mor­tals can hope for. And now J&J is hand­ing over some ex­per­i­men­tal mol­e­cules it needs to prove it’s right.

The up­start joins a long line scram­bling to ap­ply vast amounts of com­pu­ta­tion­al pow­er to­wards drug de­vel­op­ment. Their goal is to ush­er in the long-await­ed “phar­ma 2.0” and fi­nal­ly bend the ex­pen­sive curve of late-stage tri­al fail­ure. It’s un­clear how Benev­o­len­tAI’s al­go­rithms are any bet­ter at eval­u­at­ing the po­ten­tial of any small-mol­e­cule than oth­er com­pu­ta­tion­al­ly-tax­ing ap­proach­es de­vel­oped by oth­er groups — and it’s all dri­ven by the da­ta. While there’s plen­ty of sci­en­tif­ic lit­er­a­ture ripe for min­ing for new in­sight, there’s still the mat­ter of ac­tion­able and pro­pri­etary da­ta sit­ting locked away in­side com­pet­i­tive bio­phar­ma servers.

Unlock this article instantly by becoming a free subscriber.

You’ll get access to free articles each month, plus you can customize what newsletters get delivered to your inbox each week, including breaking news.

SIGN UP LOG IN BECOME A PREMIUM SUBSCRIBER