Agreed, but AI certainly makes applying easier for those who may not be confident writers.
I think it makes the quantity of applications increased. Let me use a very simple numerical answer (Obviously, there are way more candidates and firms, and people submit way more applications).
Say you have 100 candidates and 20 firms. They can write 5 apps each, so they will submit to the 10 firms they are a best fit for. For simplicity, let's say there is an equal distribution of applications to each firm. This means there are 500 applications for 20 firms, so each firm received 25 applications. They decide to take 5 candidates on, so they have a simple screening process to whittle 25 applications to 5.
With AI speeding up the research writing process, each candidate now decides to submit 20 applications. Each firm suddenly has to review 100 applications each. In addition, because of the economic outlook and AI integration, each firm is only taking 2 applicants. The process to reduce 100 applications to 2 candidates is a whole different challenge; they aren't able to use the same review system anymore.
Now imagine this, but with 5000 applicants, all with strong grades (or mitigating circumstances), 5% need reasonable adjustments (So they have to account for this when choosing the assessment method), and only 5-10 places at the end (varies a lot by firm ik). It's honestly a nightmare for grad recruitment. The good news is that most AI is not very good at writing or research at this stage. The candidates who copy/paste from AI can instantly be rejected. But those who use it roughly take ideas from it without copying output, making it difficult to differentiate from purely human-written applications. And this is where they have problems.
This isn't a law problem. It's a problem for the entire job market.