Skeptical and openly hostile scientists argued yesterday that submitting their research for censorship by intelligence agencies to prevent it from being exploited by the Soviet Union would be an unworkable nightmare and the United States would be the big loser.

Adm. Bobby R. Inman, deputy CIA director, urged scientists at the American Association for the Advancement of Science convention Wednesday to submit to censorship voluntarily because, he said, there is a "hemorrhage of the country's technology," and the Soviet military advances of recent years have been based largely on the work of U.S. scientists.

He suggested that U.S. scientists submit their work, both "prior to the start of research and prior to publication," to U.S. intelligence agencies so they can censor work considered harmful to the national security.

Yesterday, Larry Speakes, White House deputy press secretary, said: "The administration is very concerned about the loss of technology to the Soviets. It is a matter being seriously addressed by a number of departments and agencies. There is no consideration being given to any mandatory program for government review of scientific papers."

The United States will urge its allies later this month to crack down on the legal and illegal flow of militarily important technology to the Soviet Union, defense officials say.

"There have been some terrific losses," particularly in micro-electronic know-how vital to a range of modern land, sea and air weapons, said an aide who asked to remain anonymous.

One official displayed a circuit board he said was in a Soviet buoy fished out of the Atlantic by an American boatman off North Carolina about six months ago. This buoy, he said, automatically measures ocean currents and temperatures -- information valuable in anti-submarine warfare -- and radios it back to the Soviet Union. The circuits, he said, are "direct copies of U.S. circuits."

U.S. officials suggested that much of the movement of key technology through illegal channels is material that has been stolen--either by people doing it simply for money or those carrying out espionage assignments. He also said some U.S. companies assemble equipment in Third World nations and that some of their workers may make off with samples.

In attempting to deny the Soviets our best science by not publishing it, said Robert Rosenzweig, a spokesman for Stanford University, "we would lose the science ourselves. We would be the bigger loser."

He said an enormous number of scientists and their work would be involved in any attempt to shut off publication of sensitive research. Thus the program would be unworkable and "disastrous" and might lead to programs still worse to correct the situation.

William Carey, executive officer of the AAAS, the largest general science membership organization in America, said that "What alarms scientists about the Inman proposal is that once science accepts the government's right to prior restraint . . . the programs are carried out by individuals in the national security establishment. They resolve questions where there is doubt on the side of censorship rather than the freedom of scientists."

He said scientists did not want to be subject "to the whims of unknown people inside the walls of the military, not just about immediate problems, but potential ones . . . . This would be a nightmare, no more and no less than a nightmare."

Sydney Weinstein, director of the Association for Computing Machinery, said he objected to the use of scare tactics, such as talking about the Soviet threat or the threat of legislation, "to make people do what they want them to do. There should be a more rational way of dealing with this."

Carey and Frank Press, president of the National Academy of Sciences, acknowledged that there is a problem in the way technology is picked up by the Soviets and others. Press said Inman has, until now at least, opened a dialogue with the universities in a way that is unprecedented for someone in the intelligence area.

Inman already has a voluntary review system, in which the few private researchers in the mathematical theories of codes submit their papers to the National Security Agency before publication. He wants to extend it to "computer hardware and software, other electronic gear and techniques, lasers, crop projections, and manufacturing procedures."