Both! Sure: little risk when there are few or no drugs in a specific social setting, and no peers are using drugs. So it is true that social setting can be determinative in extreme cases. But if drugs are available and some peers are users then genetic variation definitely modulates risk of addiction to drugs that greatly degrade function and even mortality risk. I am sure you know last year’s excess death attributable to opiates and ilk—-now well over 100,000 in 2023— four-fold higher than in 1990. Team Sackler/Purdue Pharma, lax regulation, and greed all around, get lots of opprobrium but no jail time.

And just to admit:

Alcohol and tobacco both have much higher burdens in terms of disability-adjusted life years.

For alcohol see this plot: https://ourworldindata.org/grapher/dalys-from-alcohol-use-di...

For other drugs of abuse (excluding tobacco) see

https://www.paho.org/en/enlace/burden-drug-use-disorders