If you ask 100 people why they would prefer a mindless clone to an android, then you would likely get 100 answers. Sure some will be variations of each other, but everyone's reasons will be their own.
Part of it will be if the mindless clones can still do things that a pre-programmed AI cannot. If that spark of, for a lack of a better term, human ingenuity is there, then it is likely that the clones will still have use in things where raw numbers and calculations cut it.
At the end of the day, an AI is programmed. While it can learn, unless it is explicitly that type, it will be unable to learn things that are outside its programming.
As one example, an android can cook a mean to mechanical perfection, knowing that Ingredients A, B, and C cooked at temperature N for T minutes will make a dish. The AI may be programmed to know how to substitute certain things in the event that there is not the inventory in the kitchen. It could also just say the dish is unavailable due to a lack of ingredients.
A clone with the Chef Pack v12 could do something about it. Sure, it might not be the exact thing ordered, but the substitutions work great and were outside the scope of thing the AI would consider. A chef clone could also know the intricacies of certain dishes -- knowing how to handle them while they are in process. While yes, an AI can be programmed to pay attention during the process, their meal work is already choreographed.
Killing a clone is illegal but doesn't count as murder. You just have to pay a fee to the original owner.
Picture a movie studio shooting a film and a character has to die as part of the movie. Yes, they could use a stunt double for the filming and do what they have done for generations. But a film executive gets an idea: Why not use a clone?
While expensive, the lack of free will and the fact that the studio would likely own it means that it should be perfectly legal to do. You get to see your favourite actor bleed out with a knife in their heart and it's so realistic, you'd think it's real! But it can't be because they are still alive. Try not to think about how many deaths now are real deaths, even if they are clones.
If this becomes the norm, expect contracts to stipulate how many clones can be made from the actors and for what purposed they can be used for.
The Threshold Code
Another reason I've not seen mentioned yet is that there are people that know just enough on the topic to be wary, or they learned about AIs and androids when they were starting to emerge and not as fully stable as they are now.
While not necessarily amateur programmers, they are the people that know enough about computer code and building applications to be wary of an AI. This wariness could be built on misconceptions and incomplete knowledge.
For a parallel, consider online shopping. Many trust it and do it on a daily basis. Then there are those that don't do it because they don't trust it. Sure it can be encrypted to high heck but it won't matter to somebody that's already made a decision about it. That and the intermittent news of data breaches where information is stolen does not help.