Artificial Intelligence and the 1,200-ship Navy

This html article is produced from an uncorrected text file through optical character recognition. Prior to 1940 articles all text has been corrected, but from 1940 to the present most still remain uncorrected.  Artifacts of the scans are misspellings, out-of-context footnotes and sidebars, and other inconsistencies.  Adjacent to each text file is a PDF of the article, which accurately and fully conveys the content as it appeared in the issue.  The uncorrected text files have been included to enhance the searchability of our content, on our site and in search engines, for our membership, the research community and media organizations. We are working now to provide clean text files for the entire collection.

Developments in artificial intelligence, robotics, and other technologies could reduce drastically shipboard manning in tomorrow’s Navy. Remove the “life- support” systems and their required spaces on board ships and even a 1,200-ship Navy becomes feasible. Ironically, it is up to naval professionals to ensure that these people-saving technologies find their way to the fleet.

very naval professional must consider what s" technologies as computers, expert systems, rob^

___ ics, fiber optics, fuel cells, and composite mater*'

could mean to warship configurations in the year 2000 ^a beyond. Surely there will be a great deal more sys*^e' automation, advanced communications, dramatically *'_jt proved ocean surveillance, and high-performance com systems. If electrical superconductor development *j| proaches speculated goals, even ship propulsion and *> forms are likely to be radically different from what t are ^today-            ,_hips

It is conceivable that we will move toward wars' < with greatly decreased manning. New technology app¹'⁴^ tions surely will supplant manpower in many way^s ^ board ship, just as they are doing in factories ashore-

^^shed hard, shipboard manning might be decreased dra- _c, ^lcally, or even eliminated, while maintaining military Pability—or even increasing it. _te ^Uch developments would dramatically alter the archi- ^Ur£ of naval ships. People and their “hotel require- _w ^nts now account for most of a ship’s space and ^nt. If there were no people to support, a ship would need: berthing or messing spaces; food storage or b_a^^arat'^on facilities; potable water; a laundry, sick bay, ^ ^er shop, ship’s store, chapel, library, or other ameni­ty ’. provisions for bacteriological/chemical warfare pro- _e| ^lon^; sewage holding or processing capability; far less th_£^CtriCa' power-generating capacity and only a fraction of _Ca ^c°.^nventi°^nal heating, ventilating, and air-conditioning W-f^Cit.y' Deck spacing would not have to be seven feet. _Jt^^erfight integrity, fire containment and extinguishment, •ia!l^^enera* ^dam^ag^e resistance could be increased substan- _c, ^y- Hulls might be pressurized with inert gas to pre- ^e fires and minimize corrosion. Only minimal under- the same time. But would the nation ever countenance sending a warship to sea with little or no human control? Probably so, for two reasons. First, if there were a major threat of attack against our country, the danger of acciden­tal harm to other shipping, aircraft, or people would be accepted. Second, we have already made considerable progress along these lines with cruise missiles and inter­continental ballistic missiles. The concept of an unmanned ship is not so distant from that of an unmanned aircraft, which Tomahawk certainly is. And the Tomahawk can deliver nuclear warheads.

We have seen increasing applications of automation in our ships in recent decades, but there is nothing to suggest that we are approaching such a drastic reduction in ship- borne personnel. One key to such a development is the potential of expert systems combined with robotics. Ex­pert systems, a category of artificial intelligence, are much more than examples of sophisticated automation. The es­sentials of an expert system application are a set of rules, a

t °^re’ ^tra'ⁿ'ⁿ8 budget could be reduced in propor­tion] d°.^tTlanr|ing, and the results probably improved. Ships of be home-ported almost anywhere, with the morale Uni_t ' °^{rS anc}* their families a reduced concern. Deployed ^rotat ^W°^{Uld not} h^ave 1° t>^e scheduled for liberty ports, and to _e'°ⁿ Would be a logistic decision. Ships might not have Dio_St ^er P°rt except for periodic overhauls. Perhaps the ^Profound difference would be that a battle com­^!° emh^^Cou'^ ^use these assets without concern for danger bn_Se<j ^arked personnel; a combat risk decision would be q. °ⁿ material issues alone. ^c0s[ ?^{n SUc}h advantages, which would greatly reduce the ubl_e t° ^P’oying a warship, the United States might be ⁰ afiord a 1,200-ship Navy and reduce the budget at

knowledge base, and an “inference engine.” Only the set of rules is readily analogous to what we call automation in contemporary ship systems. We install automated devices to respond in accord with rules like: if the pressure goes up, turn the fuel down; if the water rises in the bilge, ring the bell; if the ship is off course to the left, give it some right rudder; and so on.

Some modern weapons have control modes with fairly elaborate sets of rules—the Phalanx close-in weapon sys­tem, for example. The Phalanx can be set to function auto­matically: If its radar sees a qualifying target, it fires. But many rules come into play before those bullets fly. Pha­lanx is also a good example of our peacetime reluctance to relax human control when a system can cause real dam­age. It is safe to assume that Phalanx is rarely switched to

eng¹'

The computer company people did the knowledge

its fully automatic response capability. Still, the Phalanx, for all its sophistication, is not an expert system. It has no knowledge base and cannot draw inferences from partial or incomplete data. If a target detection meets prescribed criteria, the Phalanx shoots; if the signal does not qualify, it does not. In contrast, an expert systems-controlled Pha­lanx would weigh nonquantifiable circumstances—much as a human battery-control officer would. The probability of an attack would be based on a number of qualitative factors. The most recent experience or outcome of a simi­lar target detection and engagement would be considered. The urgency of the moment—a special vulnerability or emergency damage situation on board—would bias the instant judgment of a control officer.

A closer approach to artificial intelligence can be found in modern electronic support measures (ESM) systems. Signal-intercept analyzers have knowledge bases that cata­log the characteristics of the electromagnetic signals that are likely to be encountered, both friendly and enemy. The ESM intercept system compares what it detects with its stored knowledge base. It tries to identify its contact by going through a stored set of parameters—frequency, wave form, repetition rate, etc. Again, programmed rules are followed in the analysis. The operator is advised by display about what kinds of emitters the ESM gear “thinks” it has detected. Possibilities may also be ven­tured. This is not quite artificial intelligence, but advanced ESM systems do undertake evaluation functions that would otherwise be assigned to trained and experienced personnel, and the systems may do the job more quickly and accurately than average watchstanders. Where ESM systems fall short of artificial intelligence is in applying judgment, perhaps guessing that a foreign radar might have been modified to create what appears to be a discrep­ancy, or acting on the possibility that an operator is intro­ducing unusual factors because of the current tactical situ­ation, or considering electromagnetic interference from unexpected sources. An expert system promises these kinds of extensions of automated processes. It would fur­ther the effort to bring the experience and heuristic thought processes of a human operator into the machine, which would carry out these processes with incredible quickness.

A giant step up the ladder of automation in combat sys­tems was Aegis: extensive automated self-testing features; sophisticated performance in automatically detecting, tracking, and engaging air targets; and provisions for oper­ational control modes ranging from normal console-opera­tor involvement in every function to total hands-off opera­tion. Still, Aegis is not an application of expert system technology. It was conceived before artificial intelligence became practical on a large scale, and at a time when computer capabilities were far less than those that exist today. Aegis was a pioneering achievement, exploiting the available technology. Its derivative systems will embrace artificial intelligence and other new technologies.

A current Defense Department program is attempting to develop an expert system called the “pilot’s associate” to do much of what a qualified copilot would do. For exam­ple, an aircraft pilot might depend on an expert “phan­tom” to monitor an array of interrelated flight instru­ments, make adjustments to flight systems in accordant with standard doctrine, and alert him if unacceptable con ditions exist. A reading on a given instrument may not b^e sensibly interpretable using simple rules; acceptan^ce might depend on values simultaneously reported by ^seV eral other instruments as well as parameters of the instan flight environment. A trained aviator understands these complex relationships and has developed skill in assimil^at' ing, correlating, and deciding at a glance if his avioffl^c* are functioning properly. There are reasons to believe th^a such skills can be “canned” to a safe and useful degree an expert system.

Expert systems are capable of audio communicatio¹¹' responding to spoken questions, or acting to carry out p^r0 cedures on limited oral commands. A current field off® search related to artificial intelligence involves the de^vt’ opment of “natural language” communication ^vVlt_ computers. The goal of natural language is to permit coto puters to be controlled by whatever sensible orders human might speak, write, or type. As this effort P^r° gresses, it will become less necessary to learn and folk* rigid procedures or commands in using computers. ¹ devices will become increasingly user friendly, interpw ing any reasonable human utterance correctly. The va* ^ of a natural language interface with the “pilot’s asso^cl ate” is obvious.  ..

An example of an industrial artificial intelligence apP¹ cation is illustrative. A major food-packing comP^an^J with help from a computer manufacturer, built an exp^ system to help run its soup canning cookers. The cookers multi-story, building-sized installations—handle tens thousands of cans of product at a time and do so ^ve efficiently. There are many interacting factors to be tored and controlled, and if things go awry there can great loss of production, and a lot of waste and clean-^uP As these processing installations evolved over several de ades, the company came to depend heavily on one vete ^ employee, a man who was peerless in diagnosing ¹ subtle causes of problems and determining correc¹ action before they became crises. He was well loved ^a^ well paid, but he was approaching retirement and his s pervisors realized that they could be in a lot of tr°^u without him. The solution was to formulate his knowle i base, “rules,” and inference logic into an expert sys^te .

neering and inference engine parts by working with human expert for many months, modifying their sy^st. repeatedly, and consulting countless times when uns^a factory results were encountered. The human expert coW erated enthusiastically, and now his professional cap^aD ity is immortalized.      .

It is not difficult to think of situations in which ana gous skill-capture could be valuable to the Navy. I⁰¹³® ₀- being able to clone the abilities of top-notch systems P^c pie in the fleet; to be able to install the diagnostic capah¹^ ties of the best E-6 fire-control technician you ever kn into every ship carrying the systems that he had masted < Then, when an even better systems diagnostician tur up, the expert system could be improved and revised sf ware distributed to concerned ships and support activm

ed

^r°tat°^^{an en}*‘^stment> advancement in rating, sea-to-shore

ⁱSe j.                   -       -

® ^ir>reat to your ship’s readiness.

"e_ar ; unitary departments would be among the first to Sgs _th^P? .

liven, ^{nat m}'ght be of interest to the military would ac- ^seek DoD sponsorship. If an inventor organization

^llcators, ASW people, and so on. No longer would the

ry,„ —. or just plain exhaustion of a key crew member ^ a th ^F

g_reat ^0t*^CS’ ^{a com}P^an'^on technology, appears to have /^Plication in our culture, both civilian and military. ^t()rn ^^{as alread}Y accomplished much with robotics: ^sUbm °^eS’ §^u'^ded missiles, automatically loaded guns, n_l0re^ar'^ne devices, automatic detection systems, and W_hi , An old example is the steering engine on a ship, Pilot ^{1S clearl}y ^a robotic device when coupled to an auto- $t_rj ' ^hen the sizes of naval guns and ordnance out- injQ ^ed the ability of men to handle them, robotics came |₀ P'^ay with ingenious ammunition-handling systems, WiJ devices, and power-driven gun-laying systems. d_ev , °ⁿ§er ranges, the need for better accuracy drove the fir_e °P^ment of gun directors and computers. The early <°^ntro1 computers were artificial intelligence robots, a_Cco^e °^r.^less- With springs, wires, cams, and gears they fir_e ^Phshed what skilled men had done in calculating te_tlt|^C°^ntrol values, but they did it faster, more consis- and probably more accurately.

'^vl>enK^{hines not l}‘^{re or ex}P^er>^ence survival anxiety a qq ^a 11^ets are flying. A machine can be designed to do ^c0n(p ■ ^at function well and fearlessly. It can function in he_at ^{1 l0ns} where a man cannot—in darkness, smoke, ^ ^and noise.

e,t_herⁿⁱ‘^,ny areas the Navy has been a technical leader, ^sP°Hs ^tar°^uSh its in-house research and development or by se_arch^0rin8 programs in industry and the university re- that thg °^mmunity- Until recent years, it could be assumed

of r> ■ ■                                                                                                   •

^lhiog_s P^r°mising inventions. People with ideas for new

^V% ^

could get a military contract to develop its idea, it could hope to exploit it commercially as well as sell it to the armed forces. Consequently, military planners could feel confident that American industry was working for them, that the best brains in the country were tuned to solving military problems. Their challenge was in sorting out the offerings and picking the best ideas for funding.

That situation has changed. The military is no longer the best customer for some of the technologies that may be critical to winning future wars. The Pentagon can no longer wait passively for great ideas to be delivered to Washington for approval and funding. New technologies are vital to the military, but the big dollar pay-off is in the worldwide civilian marketplace. A recent DoD-sponsored study reported that certain advanced microelectronic chips may be available only from foreign suppliers unless action is taken to foster domestic development and production.

So what can a naval officer do about all of this? What would be the point of wardroom discussions about tech­nology trends or future operational requirements if there is no way a shipboard sailor can influence such matters? Fleet people have to do their best with what they have in the way of ships, aircraft, equipment, and munitions. Sail­ors may wonder why some new things they read about, or see in the marketplace, are not available to them.

Perhaps in the back of his mind. Lieutenant Jones tells himself that he will have his chance to get some things moving for the fleet when his career takes him to Wash­ington, but meanwhile he cannot do much to influence “the system.” He will run his operation to the best of his