If they're signature based they're not better than antivirus. I have zero faith in signature based systems.
For the stuff that uses machine learning, I have to admit, I have no idea how that stuff performs. But in general I wouldn't trust a machine learning model to not be fooled.
Edit: Add to that HTTPS, I don't buy any claim that they can spot malware traffic from malware that isn't dumb, and I don't think MITMing all traffic is an acceptable solution.
You can assign 100 students to develop a trojan for a week. At the end of the week, more than 90% of the software are detected as generic trojan by the antivirus.
Probably because 90 of these 100 students have no idea how AV heuristics work and what the trivial tricks are to completely stomp them.
Don't underestimate the students and don't underestimate the AV. The world is full of surprises.
That's ok.
You're in good company with the vendors who sell ML cybersecurity appliances.
I second the IDS and WAF bullshit argument.
See http://cs.unc.edu/~fabian/course_papers/PtacekNewsham98.pdf. IDS is considered a speed bump by sophisticated attackers. Which is where you want to focus your energy.
I've been in the field for a couple of years. I work for a global corporation with 10k+ employees and most of our team members in the security department are judge on their skills and various other factors and we filter potential candidates with a small CTF. Certs have very little importance for us, but we're the exception. Most big compagnies require certifications and oddly they are the ones getting hacked.
In the field, we all know that EC-Council certs are bullshit. They are, at best, the laughing stock in infosec because their "Ethical Hacker" certification is a multiple choice answer and requires little technical knowledge and no hands-on.
However, there are a few certs out there that need a lot of work and technical knowledge to be learn for passing it, such as OSCP. It might be easy to get for someone with 10+ years but for relatively new comers, it's a really good challenge to tackle. I started with their lab, thinking it was going to be a piece of cake for me but it's more difficult than I expected, which is a good thing.
But I see your point and I mostly agree.
Care to explain why you think intrusion detection is bullshit?