It is as bad in medicine, if not worse. Recently a very influential paper about alzheimer was pulled[1] because of almost everything you said, but for biology instead: data fabrication, overestimating numbers, etc.

There's also a case related with cloning (which includes stem cell research) and massive fraud in South Korea 20 or so years ago.[2]

It's a complicated matter: research is a beacon of light on top of a mountain of failed experiments. And not even the government will fund a mountain of failures if your shining discovery at the end can't bring returns to society (or shareholders) in greater value than what you took.

And the pharmaceutical industry never settled for small results. And, unlike ML, it's much harder to test things as an outsider (you can hate the hype, but openAI, hugging face and all these "open" language models makes it much easier to learn, test, tweak and improve things without a master's degree in data engineering.).

It's a grimy area to research even as a pastime. There's a lot of recorded practices both in research and production that makes you question how low can a person go for money

1: https://www.science.org/content/article/potential-fabricatio...

2: https://youtu.be/ett_8wLJ87U