I generally have a problem when people say ML algorithms are either 1) blackbox or 2) unexplainable. ML algorithms are algorithms and they are made by fellow humans in a very structured formal method called programs. These programs are completely transparent and completely explainable. If you wanted to visualize a neural network, you can do so even to the bit level.
People have no problem believing another human who is much more complex and much more prone to many different factors like emotion, weather, culture, financial rewards, etc. And if you were to tell me that you cannot trust an ML algorithm because you don't understand how it works, you need to think again about what you're saying. YOU DON'T UNDERSTAND HOW HUMAN BRAIN WORKS EITHER.
The algorithms are not blackbox, the way companies use them are blackbox. Amazon's recommendation algorithm is 100% transparent, it's just that they won't tell you how they're using the algorithm, what they're feeding it to achieve what. What you can't trust isn't the algorithm, it's the company/service/product. You just can't trust another human (or group of humans). Yet, you "believe" certain politicians will make your life better.
It's a more profound problem than just not understanding how things work. But I have this one speculation.
You don't understand exactly how every part of an automobile works. But you use it everyday. You've learned combustion engines, and laws of physics in school. So you're more or less comfortable, don't have a problem with driving cars and riding buses.
Most of us haven't learned probabilistic thinking and computational thinking, and I feel like that's where the fear of "trust" comes from. So I think our education systems should reflect the need for understanding the basics of probabilistic computational thinking.
Would be interested to dig deeper and find out more.