With JSONB I can receive a boolean element with five states: null, true, false, invalid (i.e. actually a string, number or array value), or simply nonexistent, and it's up to the application to deal with all such cases. You may have to treat the database as a source of potentially invalid data that must be sanitized.
With regular DB columns you can reasonably assume that a boolean not null column will a) exist for all records and b) return either true or false. You need only scan the database schema at application start to verify this.
On the other hand, if you're using JSONB for an options structure then the possibility of key nonexistence may actually be useful, since it implies "use system default". (Yes this can differ from the meaning of a NULL value, especially if you are merging options structures).
You might question whether it is wise to have a "boolean" key/value pair with four semantically distinct states. I can only say I have done this and I am not proud of it.
This is an anti-pattern because you can't model "1:1" that way. Instead, it will be "1:0..1", and now you have some nasty corner cases when the first table has an entry whose counterpart in the second table is missing. Also, when using a column now you always have to think about which table had it - the first or the second table?
I had to work with such a design in a real-world project and it was really annoying.
This may make sense if you have custom fields, i.e. the set of keys is user-defined and ever increasing.
Apart from that, there is no reason to do this. If it is all about avoiding having "too many" columns and/or "too may" null values, then I'd say: Don't worry. Just use as many columns as you need.
I would even go one step further and say: It is a common anti-pattern to introduce generic key-value stores (through a separate table, JSON structures or XML structures) without a compelling reason.