Prime numbers have extremely well understood patterns and this is what he's seeing. There's a weird and persistent myth that there's 'no patterns in prime numbers' but of course factors repeat at known intervals and prime numbers are the inverse of numbers with factors. So if you can accept that numbers with factors have a pattern to them (which should be obvious, they repeat at known intervals by definition) you should be able to accept prime numbers, the inverse of numbers with factors, have patterns too since they are just the gaps in the pattern of numbers with factors.

These patterns were documented and well understood starting in BC times by Erasthosenes and learning them as part of prime number theory is a 101 course in tertiary maths education. So it's really really weird for anyone to say "there's no patterns". There are and they are extremely well understood and known.

Here's a simple pattern; All prime numbers above 2 are odd. Well duh right? Otherwise they'd be a multiple of 2, not prime.

Well let's extend this. All prime numbers above 6 are of the form 6n + 1 or 6n +5. Otherwise they'd be a multiple of 2 or 3.

Once more; All prime numbers above 30 are of the form 30n + one of [1,7,11,13,17,19,23,29]. Anything else would be a multiple of 2,3 or 5. You can extend this forever. Note each time we do this we're reducing how many numbers could possibly be prime. From 1/2 to 2/6 to 8/30 numbers possibly being prime. Keep going with this and you'll converge to the prime counting function.

Basically whenever you have a composite number there's well understood periodic gaps in primality. People understand this more intuitively for base 10 where anything ending in 0,2,4,6,8 is a multiple of 2 and anything ending in 0,5 is a multiple of 5 hence you only get primes ending in 1,3,7,9 when writing in base 10 but this idea works for any composite number. This leads to the extremely well known and well understood patterns you get when you graph primes in various ways.