He simply says that most of the time senses--for what information they do provide--provide useful/accurate information.
In his 2015 book he calls the Argument from Illusion and variations thereof the Bad Argument. He simply says that because someone who experiences a hallucination really sees nothing real, it doesn't at all imply that most people's vision works like that in usual circumstances. So concluding that there's no connection between vision reality in general is a bad argument.
He says/concludes e.g.
I have to say that what bothers me about Phenomenalism and the Representative Theory is not some technical problem that they have, but their sheer preposterousness.
Although Searle doesn't do this as far as I can tell, one can obviously invoke e.g. evolutionary mechanisms: if senses were completely misleading and useless they may not have arisen or survived this long etc. There are such works elsewhere.
(In general I find philosophy works about vision seriously underwhelming. Even when they come to a biologically plausible conclusion, they seldom bring the right arguments.)
Your [counter]argument is that blind people exist. Sure, but most people are not blind. So you can't say itself blindness evolved for some use/purpose in people, as opposed to being an accident/mutation. (That vision is not that useful for moles as opposed to people is another matter--different environmental pressures.) That blind people can still deal with the world with the remaining senses isn't any sort of refutation that blindness is not the norm for people, nor that vision is not plausibly useful as a tool for perceiving reality.
Likewise, the fact that the senses we normally have don't provide complete information like say x-ray vision isn't an argument that those sense we have are always misleading.
Sure, you (and other philosophers) may disagree with this, but that's basically what Searle is saying, although I've improved on his account quite a bit here...
If you want to argue that whole world is a simulation, then sure, the above account only argues that vision is a good adaptive mechanism within that (simulated) domain. But most realists would also deny that the world is a simulation.