One problem with cracking Linear A is, there are no known multilingual inscriptions, or anything that may have been written on transitory materials. The second problem is, there aren’t a lot of tablets of Linear A, and a larger sample would be needed to try this technique.
A number of people (including Barry Fell) have claimed to have cracked it, but neither the people who have made the claims nor linguists agree that anyone has done it. :^)
I like Barry Fell’s books. He seems to stretch his conclusions a bit beyond safe reasoning sometimes but I think he is basically sound. I have been to see the structures in New Hampshire. Translating linear A by AI might be like climate “models” and depend on the assumptions put in.