The thrust of the Launch Escape System (LES) might look significant, but the total impulse is not (that is, they burn for such a short time, they can't actually impart much delta-V, especially not while the spacecraft is still attached to its mostly-fueled carrier rocket). The longer the LES remains attached to the spacecraft, the more impulse the carrier rocket wastes accelerating the mass of the LES toward orbital velocity.
Now, yes, there's probably an optimal point - perhaps after first stage separation, so the carrier rocket mass is lower - when firing the LES with it still attached to the spacecraft, and then jettisoning the LES (perhaps with the last of its thrust), would produce some benefit. However, this is probably less safe than just jettisoning it cleanly. The LES is an emergency system; it's as safe as it can be, of course, but its design mandates instant, powerful, and reliable operation; those values all trade off, to some extent, against safe operation. If the LES renders the spacecraft un-spaceworthy, that's acceptable for its intended use - that spacecraft is having a bad time and will not go to space today anyhow - but would obviously be unacceptable as part of the normal launch process.
For that matter, the interface between the spacecraft and upper stage of the carrier rocket might not be designed to tolerate tension at all (after all, in normal operation, it will only experience compression or free-fall), and the LES is, by its design, trying to accelerate the spacecraft more than the booster is accelerating it, so there will be tension[*] (in theory, any scenario that triggers the LES should also trigger MECO, but you can't assume the main engines are under control in that kind of scenario either, so the LES has to be able to pull the spacecraft away even under full burn of the booster).
Given those risks, it seems decidedly not worth trying to get a few extra m/s (when you factor in the mass of the carrier rocket) by firing the LES as part of a standard ascent process.
[*] As @TooTea points out, quite a lot of tension, actually. If the LES generates 14g of acceleration, while the carrier rocket is generating at most 4g, that's going to be a massive and abrupt 10g of tension shock hitting the payload interface. Definitely not something you want to have to account for in normal operation! Even if it's possible to do safely without impairing the LES in abort scenarios, any mass you add to that interface to beef it up for that comes directly out of your payload mass budget. All for a few hundred thousand (at best) lb*s of impulse, lost in the noise of what the carrier rockets provides.