ChatGPT became the validation of transformer models and generative AI. A relatively nameless validation is also underway in the realm of drug discovery with less fanfare but potentially equally meaningful implications.
https://www.drugdiscoverytrends.com/generative-ai-impact-drug-discovery/
When one skims articles like https://bit.ly/3NzYBhr or peruses research papers like https://bit.ly/447Ss30, it is easy to be convinced of the role generative AI could play in drug discovery. However, just as tech behemoths such as Google and Microsoft needed tangible evidence of the efficacy of these models to start pouring investments into this domain, so does the drug discovery industry need concrete validation.
Transformer models and generative AIs have been in innovation circle discussions for at least three-four years, but they were not attracting the senior management's attention and big Dollar investments until the validation provided by ChatGPT (https://bit.ly/45QZrhZ provided one explanation of what changed last year). These technologies' transformative nature is not a recent development, but unlocking their full potential necessitates a financial catalyst.
INS018_055 from Insilico might not be as catchy a name as ChatGPT. Being the first AI-designed novel molecule that has now advanced to a Phase 2 trial, its progress could prove pivotal for AI-driven drug discovery worldwide. It represents a success story that could bolster faith in AI for drug discovery. This field is seen as a theoretical possibility and may turn into a tangible reality with investments and attention from all involved.
Similar proofs of concepts have to emerge if not already emerging, in other sectors like robotics and automobiles, to name two. Validations must also emerge at country levels as the initial force for the metaphorical flywheels to start spinning. GenAI revolution is dominated by American companies so far, with the rest not as razor-focussed (a good article on India as an example from my friend Andy Mukherjee - https://bloom.bg/3JE9kGD), but this too is almost bound to change as convictions levels rise.
