JARRETT: When it comes to telling the truth, it seems lie detectors flunk the test. That's right, this according to the panel of scientists who spent 19 long months studying polygraph machines. They found that lie detectors often say that people telling truth are lying, and vice versa. Still, some experts stand behind the accuracy of polygraphs. A handful of government agencies still use them quite a bit.

For more on this controversy surrounding so-called lie detector tests, we're joined now from San Francisco by Professor David Faigman of the University of California, San Francisco, Hastings College of the Law, my alma mater. Say hello to Dean Kane for me. And with us from Atlanta: Dan Sosnowski, who is the director of the American Polygraph Association.

OK, Dan, this is the esteemed National Academy of Sciences that is now saying you can get away with lying to a lie detector test and the test itself can lie about results by falsely suggesting an honest person is lying. You're the pro. Is the academy wrong?

DAN SOSNOWSKI, AMERICAN POLYGRAPH ASSN.: Well, it's nice that we are finally get some research in this area of screening.

I tend to agree with the academy saying that there is a lack of research showing that the polygraph is an absolute scientific tool in that particular area. But, obviously, the results that we come by with the polygraph is great, because individuals who are coming in are telling us constantly information that they would never divulge if it wasn't for the polygraph.

JARRETT: Well, but that doesn't mean that the test itself is reliable, but it may be a good coercive technique. Look, they spent 19 months, Dan, studying polygraphs. They came to the conclusion it's not reliable enough to use in employee security screening for federal agencies. Would you use it for that purpose?

SOSNOWSKI: Oh, absolutely. We have to look at what's being done and what kind of questions are being asked and what's the end result. Are individuals coming in there-not just as a coercive tool. If individuals clearly fail a polygraph-and that's by having physiological reactions observed on the test-we afford them the opportunity to say what was bothering them. And they end up telling us a lot of information.

JARRETT: Professor, what do you think about the accuracy of polygraphs?

DAVID FAIGMAN, HASTINGS SCHOOL OF LAW: Well, let me be very clear. The committee said that it was not valid, in that it's a very weak tool, something more like an ax than a scalpel.

And so what we're saying really is that the Department of Energy, the Department of Defense, the FBI, the CIA, should not rely on the polygraph and certainly not rely on the polygraph to the exclusion of other technologies. The validity, the accuracy is extremely weak. It's likely to produce many, many false positives, people accused of being spies when in fact they're not spies or terrorists, or they're not terrorists.

And, perhaps even more troubling, it's likely to allow some people who are indeed spies or terrorists to get through.

(CROSSTALK)

JARRETT: Would you recommend that it be eliminated completely and used for absolutely nothing?

FAIGMAN: Well, we are not necessarily saying that it should not be use for utilitarian purpose. As you said yourself, it may indeed be a very effective interrogation tool.

But, of course, as long as the examinee believes that a refrigerator or a Xerox machine is a very effective tool, then people will come in. And if they're told they're not doing very well on the test, then they might very well own up to minor security violations. Or, as we see in the criminal context, they might own up to criminal activity.

JARRETT: Dan, you want to respond to that?

SOSNOWSKI: Well, again, it's interesting where the studies are coming from and why are they basically saying it doesn't work.

And it's just like if an individual was requested to submit to a drug test-and sometimes they'll say, "Let's go for a 10-panel type of a screen," that means they're looking at 10 different type of potential drugs that's used. If the individual comes back positive, it doesn't mean he's using every single drug. We have to look at in a screening aspect vs. a very specific type of a situation.

JARRETT: But would you disagree with what the professor said, when the professor said, in terms of accuracy, it's extremely weak.

SOSNOWSKI: Well, just a stand-alone, where you're going to look just only at reactions, possibly it is somewhat weak. But we have to look at the whole totality of that picture.

JARRETT: Gentlemen, I'm so sorry. We're out of time. I hope you'll come back.

Professor David Faigman, Dan Sosnowski, thank you very much.

FAIGMAN: My pleasure.

SOSNOWSKI: Thank you.