<Hmm only a few seconds, huh?
Well if they could be designed to fire for about 20 seconds, then perhaps they could always be lit just after main ignition. This way they could add to the thrust at blastoff and be used as emergency escape. This means that after about the 18th second they would be jettisoned.

First, the escape system was designed to be useful for several minutes of flight, not just the first few seconds. If memory serves, the Apollo LES was jettisoned well into the 2nd stage burn. The LES was designed to pull the entire capsule free of the rocket from the launch pad or up to a certain altitude, say 100,000 feet. According to this source, the LES was jettisoned 3 1/2 minutes into the flight.

With your idea, once the LES rocket is almost out of fuel, you've lost any hope for escape.

The Apollo LES used a solid-rocket motor with 155,000 pounds of thrust. The motor was designed to burn for a few seconds, sufficient to get the capsule clear of the rocket stack and deploy the parachutes. Solid rocket motors typically have a specific impulse in the 200 range, so each second of burn at 155,000 pounds of thrust and an Isp of say 220 seconds means the rocket needs 704 pounds of propellant. Increasing the burn time also increases the weight of the LES.

There's also another problem - what happens if you have to do a pad abort immediately? Solid rocket motors are hard to shut down once lit and pulling high gees for 20 seconds isn't fun. With a Command Module mass of 5,806 kg, the LES would've subjected the crew to over 12 Gs acceleration. Even lying down, that would hurt.

From Astronautix.com

29 June 1965 Apollo PA-2 launch escape system test Program: Apollo.
NASA launched Apollo mission PA-2, a test of the launch escape system (LES) simulating a pad abort at WSMR. All test objectives were met. The escape rocket lifted the spacecraft (boilerplate 23A) more than 1,524 m (5,000 ft) above the pad. The earth landing system functioned normally, lowering the vehicle back to earth. This flight was similar to the first pad abort test on November 7, 1963, except for the addition of canards to the LES (to orient the spacecraft blunt end forward after engine burnout) and a boost protective cover on the CM. PA-2 was the fifth of six scheduled flights to prove out the LES.