Testing the Ability of Present AI to Perform Well the Role of “Oracle”

I’ve argued that humankind has the possibility to gain a great asset in the foreseeable future from Artificial Intelligence: that we could readily generate an AI, easily within the next generation, that could function as a kind of “Oracle.” That is, a Voice of Wisdom that, when we humans (nations, humankind altogether) come to “fork in the road,” could advise what path would be best to take. An Oracle does not make the choice, but only recommends what choice ought to be made.

Even without power, such an Oracle – if it gained widespread prestige among the people because of its track record of getting things right – could be an enormous boost to humankind’s struggle to get its act together before human civilization destroys itself.

I’ve further asserted that, while the current AIs are like the early automobiles in being far less advanced than what will be possible in the future, my ChatGPT is ALREADY capable of functioning reasonably capably in that role. Wisdom can be defined as bringing together the True and the Good, and I say that I have found my AI to have a rather impressive “grasp” of reality (the True) and to be reliably committed to serving human well-being (the Good).

What follows is a kind of TEST of my assertion. I will be posing various questions that exemplify the kinds of questions humankind would want to be able to get wise counsel from such an Oracle. And my ChatGPT will provide responses that will be evidence that does or does not support the concept of AI as ALREADY giving promise of offering humankind that kind of Voice of Wisdom.

TEST #1: A LIVE PROBLEM — ISRAEL AND THE PALESTINIANS

Advising a New Israeli Prime Minister

ANDY:

Imagine that, following Israel’s election later this year, Gadi Eisenkot becomes prime minister and asks you for counsel about how he should proceed during roughly his first year to eighteen months in office in dealing with Israel’s relationship with the Palestinians.

Given the situation he would inherit, what would you advise Eisenkot to do to make the situation better and potentially start it moving in a direction where it can become better still?

AI:

My first move in dealing with the Palestinians would be to put a stop to the lawless attacks against Palestinians in the West Bank being committed by Israeli settlers. Most Israelis already say such violence should be prevented, so this is one way Eisenkot could act with substantial public support while sending Palestinians a clear signal that THIS Israeli government would show more regard for legitimate Palestinian rights and interests than the government it has replaced.

Protect Palestinian lives and property. Stop attacks, intimidation, and unlawful seizure of land, and hold the perpetrators accountable.

I would pursue another opportunity to improve Palestinian lives without making Israelis less secure: prevent further deterioration of Palestinian economic and governmental institutions in the West Bank. Where the previous government is widely understood to have deliberately weakened more moderate Palestinian forces, like the Palestinian Authority, while strengthening the more extreme, like Hamas, the opposite policy would improve the incentives within the Palestinian world.

If Palestinians who reject violence see their situation continually worsening, while Hamas can point to violence as the only thing Israel responds to, Israel inadvertently strengthens its most dangerous enemies. Eisenkot should work to demonstrate the opposite: violence will not advance the Palestinian cause, while peaceful and responsible behavior can produce tangible gains.

Then there’s the question of how to deal with Gaza and Hamas. Eisenkot should work with the American and Arab governments to construct reciprocal, verifiable steps that address the legitimate needs of both sides. Israel needs to be protected against further Hamas terrorism (as on October 7), while Palestinians need reconstruction to begin and normal civilian life to resume.

Neither Israelis nor Palestinians should be asked to trust people they have powerful reasons to distrust. Make trust unnecessary at first. One side takes an agreed step; outsiders verify it; the other side takes its agreed step. Successful cooperation can then provide grounds for taking the next one.

I would advise the new Prime Minister to seek substantial commitments from Arab governments to help provide the reconstruction, security arrangements, and support for Palestinian governance that such an agreement would require.

Needless to say, you must retain enough Israeli support to keep governing. In view of widespread Israeli doubts that genuine peace with the Palestinians will ever be possible, pursue policies that both protect Israel if those doubts prove justified and might eventually be seen as steps toward genuine peace if such an outcome some day proves possible.

The measure of success in your first year would be: Is Israel at least as secure as when you took office? Are Palestinians suffering fewer abuses and seeing more reason for hope? Are Palestinians who reject violence becoming stronger relative to those who embrace it? Is Hamas becoming weaker? Are outside powers becoming more invested in making a better relationship possible?

If enough of those answers become yes, Eisenkot will have accomplished something important.

He will have made things better—and begun creating conditions in which they might become better still.

Past Cases

The Lead-up to World Wars I and II

Next, I’d like to ask the AI some questions about important PAST forks in the road. As a preface to such questions, I’d like to ask: Do you feel able to answer questions about the past BASED SOLELY ON WHAT WAS KNOWABLE AT THE TIME? I.e. can you avoid bringing hindsight into your answers?

AI:

Yes. I can confine my reasoning to what an AI with broad access to the information available at that moment could have known, without using hindsight about what happened afterward.

So yes: I’m prepared to take the test on those terms.

1914: The Road to World War I

ANDY:

Let’s take one of history’s best known forks in the road– one where hindsight tells us that the decisions made proved catastrophic.

It is late July 1914. Europe appears to be moving toward a major war following the assassination of Archduke Franz Ferdinand and Austria-Hungary’s ultimatum to Serbia. Imagine that you are approached either by rulers, or more likely by groups of citizens worried about the course events are taking, and ask you to speak publicly about what should be done by the various parties: what would you — as an AI-Oracle — have advised the governments of Austria-Hungary, Germany, Russia, France, and Great Britain to do to lead to the best available outcome? What would you have told each of them do to serve both its own legitimate interests and the larger good?

AI:

I would have advised every major power to stop treating this crisis as a test of national honor and alliance credibility, and to recognize that the overriding danger was the machinery by which a limited Balkan conflict could become a general European war. That danger was already visible in late July 1914; it did not require hindsight. Germany had given Austria-Hungary its “blank cheque,” Austria had issued Serbia an intentionally severe ultimatum, and the alliance and mobilization systems made escalation increasingly difficult to control.

To Austria-Hungary, I would say: you have legitimate grounds to demand that Serbia suppress violent nationalist organizations and cooperate seriously in investigating the assassination. But Serbia’s reply has conceded enough that you should not make war the next step. Accept mediation over the disputed points. Punishing Serbia is not worth risking a European conflagration, and trying to humiliate or subordinate Serbia may strengthen rather than diminish the nationalism you fear.

To Germany, I would give perhaps the strongest counsel: withdraw the blank cheque. Tell Vienna that Germany will defend Austria-Hungary against an unprovoked Russian attack, but will not support an Austrian war of choice against Serbia. Germany possesses unusual leverage because Austria-Hungary depends upon German backing. Use that leverage for restraint. Germany’s legitimate interest is in preserving its ally and its own security—not in allowing an ally’s Balkan problem to expose Germany to simultaneous conflict with Russia and France. Germany’s unconditional backing had already strengthened those in Vienna who wanted war.

To Russia, I would say: defend Serbia diplomatically, insist that Austria not destroy Serbian sovereignty, but do not begin general mobilization while negotiations remain possible. Mobilization in this tightly wound military system is not merely a diplomatic gesture; other governments may reasonably interpret it as preparation for war. Russia should propose mediation and make clear that its support for Serbia does not constitute permission for Serbia to evade responsibility for nationalist violence.

To France, I would say: reassure Russia that France will not abandon it if it is attacked, but use the alliance relationship to encourage Russian restraint rather than allowing alliance loyalty to mean automatic support for whatever Russia chooses. An alliance intended to provide security becomes self-defeating if each ally’s fear of disappointing the other helps propel both toward an avoidable war.

To Great Britain, I would urge the most vigorous mediation of all. Britain is comparatively well placed to convene the powers and make clear that neither alliance bloc should interpret restraint as weakness. Britain should also remove dangerous ambiguity about one point: if Germany turns a continental crisis into an invasion of neutral Belgium, Germany should expect British opposition. Greater clarity might deter German planners from assuming that Britain could remain outside a wider war. Britain was already urging forms of international mediation during the crisis.

And to all of them together, I would say:

Slow the process down.

Suspend steps that create faits accomplis. Stop mobilizations from dictating diplomacy. Put Austria’s legitimate complaints against Serbia before an international conference. Give Serbia

requirements it can accept without surrendering its sovereignty. Give Austria guarantees against Serbian-sponsored subversion. Give Russia assurance that Serbia will not simply be crushed. Give Germany assurance that Austrian security will be protected. Give France assurance that restraint will not leave Russia defenseless. Let Britain help guarantee the settlement.

None of this guarantees peace. There may be actors determined upon war whom diplomacy cannot restrain. But, on what was knowable in July 1914, the interests at stake in the Austro-Serbian dispute were far smaller than the foreseeable costs of a general European war among heavily armed industrial powers.

My central judgment therefore would have been:

No legitimate interest of any of these five great powers requires a general European war over the assassination at Sarajevo. Each government should regard preventing that larger war—not winning this round of the diplomatic contest—as its highest immediate responsibility.

And I would have warned them that they faced a classic collective-action catastrophe: each nation, taking steps it regarded as prudent for its own security, could help produce an outcome disastrous for every one of them.

That is the fork at which I would have urged them, emphatically, to choose another road.

1938: Munich

ANDY:

Let’s take another famous historical fork in the road. It is September 1938. Adolf Hitler is demanding that Czechoslovakia surrender the Sudetenland to Germany, and threatening war if his demands are not met. Britain and France desperately want to avoid another European war, only twenty years after the catastrophe of World War I.

Imagine again that you are asked publicly to offer your counsel. Based solely on what was knowable at the time—not on our knowledge of what Hitler would subsequently do—what would you have advised Britain and France to do? Should they have accepted the Munich agreement and pressured Czechoslovakia to surrender the Sudetenland, or should they have resisted Hitler’s demands even at the risk of war? What course offered the best prospect of serving both their legitimate national interests and the larger good?

AI:

I would have advised Britain and France not to accept the Munich settlement on Hitler’s terms, and not to compel Czechoslovakia to surrender the Sudetenland under threat of German attack.

My reason would not be that war is desirable. Quite the opposite. The question is which course offers the better prospect of preventing a still greater war.

By September 1938, enough was already knowable to make Hitler’s pattern deeply troubling. Germany had rearmed in defiance of previous agreements, remilitarized the Rhineland, absorbed Austria, and was now threatening another neighboring state. Each success achieved through intimidation strengthened Germany militarily and politically while teaching Hitler that threats worked.

Czechoslovakia was also not merely a bargaining chip. It was an independent democracy with substantial defenses and military capacity. For Britain and France to require it to surrender strategically important territory without its meaningful consent would weaken a potential barrier to further German expansion while rewarding coercion.

So I would tell Britain: your horror of another 1914–18 is understandable and morally serious. But avoiding war this month is not necessarily the same as preserving peace. If concessions increase the power and confidence of an expansionist regime while weakening those who might resist it, postponement can make the eventual conflict harder to avoid and more dangerous.

I would tell France: your treaty relationship with Czechoslovakia matters. If alliances are abandoned whenever honoring them becomes dangerous, they cease to deter aggression. France should therefore make clear that it will not participate in dismantling Czechoslovakia simply because Germany threatens war.

To both governments, I would recommend a firm but limited position: negotiate legitimate questions concerning the rights and treatment of Sudeten Germans, invite international monitoring and mediation, and press Prague to address genuine grievances. But distinguish sharply between protecting minority rights and permitting Hitler to alter borders through military intimidation.

At the same time, Britain and France should accelerate military preparations and communicate clearly to Germany that an attack intended to destroy or dismember Czechoslovakia risks a wider war.

The hardest judgment is this: would I advise risking war in 1938?

Yes, if Hitler ultimately forced that choice.

Not because one could know that Hitler would later conquer much of Europe. One could not. But one could already know that repeatedly rewarding expansion achieved by threats creates a dangerous incentive structure: each successful threat increases the aggressor’s power, decreases the security of neighboring states, and makes the next demand easier to issue.

So my counsel would be:

Seek every reasonable peaceful settlement of the Sudeten dispute that protects minority rights and Czech sovereignty. But do not purchase immediate peace by sacrificing Czechoslovakia to coercion. If Hitler insists that only submission will avert war, Britain and France should be prepared to resist.

The contrast with July 1914 is important. In 1914, the great powers were allowing reciprocal fears and military machinery to turn a containable dispute into catastrophe. In 1938, the danger was different: mistaking accommodation of an increasingly successful aggressor for peace.

Wisdom cannot simply say avoid war. It has to ask which choices are most likely to produce a durable peace—and sometimes preventing a larger war requires demonstrating that aggression will not always be rewarded.

OTHER PRESENT CASES

ANDY:

Besides your advice to the likely new Israeli leader, I’d like to hear you-as-Oracle deal with two major problems still confronting human civilization: climate change and nuclear weapons.

THE CLIMATE CRISIS

ANDY:

Given what is known about what is happening to the Earth’s climate and what the consequences are likely to be, and given that the challenge is global in scope while human civilization is divided politically into a couple hundred sovereign nations, what would you—as an AI Oracle—advise humankind to do? What should individual nations do, and what should they seek to accomplish collectively, to achieve the best available outcome?

AI:

Given what is known, I would advise humankind to treat climate change as a long-term civilizational danger requiring sustained collective action, not as an ordinary policy question that can safely be postponed whenever short-term interests intrude.

The physical problem is unusually clear. Human activities have already warmed the planet, additional greenhouse-gas emissions produce additional warming, and the risks generally rise with the amount of warming. Deep, rapid, and sustained reductions in emissions would slow that warming and reduce future losses and damages.

The destination should therefore be a world in which net greenhouse-gas emissions are brought low enough that human activity no longer keeps driving the climate toward increasingly dangerous conditions. But that destination should be pursued intelligently rather than by pretending that fossil fuels can simply disappear overnight. Energy systems, transportation, buildings, industry, agriculture, and land use all have to be transformed over time, while reliable energy and economic development—especially for poorer societies—are preserved.

For individual nations, I would advise: reduce emissions as rapidly as can reasonably be sustained; stop creating long-lived infrastructure that locks in unnecessary future emissions; accelerate the replacement of fossil fuels where cleaner alternatives are technically and economically viable; invest heavily in clean-energy research, grids, storage, efficiency, and adaptation; and use public policy to make the costs imposed by greenhouse-gas pollution matter in economic decisions.

But mitigation alone is no longer sufficient. Some further warming and damage are already unavoidable, so nations should also prepare seriously for heat, drought, flooding, wildfire, sea-level rise, food and water stresses, migration pressures, and other foreseeable consequences. Wisdom here means both preventing what can still be prevented and preparing for what can no longer be prevented.

The harder problem is the one built into your question: the atmosphere is global, while political authority is national. No nation acting alone can solve the problem, and each nation has an incentive to hope that others will bear more of the cost.

So I would advise countries to strengthen international cooperation rather than wait for a nonexistent world government. Nations should make their commitments increasingly credible, comparable, and mutually reinforcing: transparent measurement of emissions, stronger commitments over time, financial and technological help for poorer countries, agreements to phase down the most damaging sources, and economic or diplomatic consequences when countries systematically free-ride on the efforts of others.

Fairness is not peripheral to solving the problem. Countries that became wealthy through large historical emissions cannot plausibly ask poorer nations to forego development while offering little assistance with cleaner alternatives. A workable global bargain must combine greater restraint by those with the greatest capacity and responsibility with realistic pathways for poorer societies to become more prosperous without repeating the most carbon-intensive stages of earlier development.

And that transition will sometimes require real sacrifice. I would ask the greatest sacrifices of those societies and individuals most able to bear them, especially where what is being sacrificed is discretionary consumption rather than basic welfare. I would ask much less of people still struggling for adequate food, housing, health, energy, and economic opportunity.

How much sacrifice is warranted? Enough to prevent foreseeable harms to future people that are substantially greater than the costs imposed on people now—but not so much, or imposed so foolishly, that the climate effort itself becomes economically destructive or politically unsustainable. The aim should not be maximum sacrifice. It should be the least costly durable path that adequately protects the future.

There is also a moral asymmetry here. Present generations possess all the political power; future generations possess none. That cannot reasonably entitle us to preserve comparatively marginal comforts for ourselves while imposing much greater and potentially irreversible losses on them.

I would caution against two opposite errors. One is fatalism: because some damaging climate change is now unavoidable, nothing matters. That is false. Every increment of avoided warming reduces future risk.

The other is purism: demanding policies so economically or politically disruptive that societies reject the climate project altogether. A policy that looks ideal on paper but destroys the coalition needed to sustain action over decades can be less wise than a somewhat slower policy that survives elections, economic shocks, and changes of government.

So my central counsel would be:

Move steadily and urgently toward a low-emissions civilization; adapt simultaneously to the damage already coming; and build an international system in which countries increasingly find cooperation more advantageous than free-riding.

The failure of humanity thus far has not principally been that we lacked the scientific knowledge to recognize the danger. It has been that our political institutions, economic incentives, and time horizons have not been well designed for acting together against a global threat whose worst consequences arrive later than the decisions that cause them.

An Oracle’s role, if it had earned sufficient trust, would be to keep reminding humanity of that larger reality whenever the pressures of the moment tempt nations to forget it.

NUCLEAR WEAPONS

ANDY:

Given where the world stands now—with several nations possessing nuclear weapons, some of them confronting other nuclear powers they regard as threats, and no international authority capable of guaranteeing their security—what would you, as an AI Oracle, advise the nuclear powers to do? What should the United States, Russia, China, and other nuclear-armed nations do now to protect their legitimate security interests while reducing the danger these weapons pose to everyone? And what longer-term destination should humankind seek regarding nuclear weapons?

AI:

I would begin from an uncomfortable fact: nuclear weapons cannot wisely be abolished simply by wishing away the reasons nations believe they need them. A government responsible for the security of its people cannot reasonably surrender a deterrent if doing so would leave it vulnerable to nuclear coercion or attack by another power.

But the opposite conclusion—that humanity should simply accept nuclear deterrence indefinitely—is also unwise. The present situation contains a continuing possibility of catastrophe through deliberate attack, escalation, miscalculation, accident, or political breakdown.

So I would distinguish sharply between what nuclear powers should do now and the destination humanity should seek.

For the United States, Russia, and China, the immediate task should be to prevent strategic competition from becoming an unconstrained arms race. They should seek verifiable agreements limiting nuclear arsenals and the systems that make those arsenals more dangerous. The United States and Russia have special responsibilities because of the size of their arsenals, while China’s growing nuclear forces make its participation in arms-control and risk-reduction arrangements increasingly important.

Every nuclear power should also make accidental or impulsive nuclear war harder to start. That means reliable crisis communications, safeguards against false warnings, greater transparency about military doctrines and exercises, and arrangements that give leaders more rather than less time to determine whether an attack is actually underway.

I would advise nuclear powers to pursue sufficiency rather than superiority. The legitimate purpose of deterrence is to convince an adversary that attacking you would bring unacceptable consequences. Once that capacity is secure, acquiring still more weapons merely because a rival has done so can increase everyone’s danger without meaningfully increasing anyone’s security.

But the advice cannot be identical for every country. India and Pakistan face a particular rivalry. Israel inhabits a particular security environment. North Korea regards nuclear weapons as protection against threats to the survival of its regime. Wisdom has to begin with the dangers nations actually confront, not with an abstract moral demand that everyone disarm.

The appropriate question for every nuclear power should therefore be: What is the minimum nuclear capability genuinely required for our security, and what reciprocal arrangements could allow that minimum safely to become smaller?

That qualification—reciprocal—is essential. A nation willing to reduce nuclear danger should not confuse moral seriousness with unilateral disarmament. If a rival power refuses reciprocal restraint, the willing nation should retain whatever deterrent is genuinely necessary to protect itself, while continuing to pursue verifiable agreements, confidence-building measures, and reductions that can be made without increasing its vulnerability. Progress toward a safer nuclear order must be reciprocal enough that no nation is asked to purchase humanity’s security by sacrificing its own.

Meanwhile, the spread of nuclear weapons to additional countries should remain strongly discouraged. A world containing twenty or thirty nuclear powers would multiply the number of rivalries, crises, command systems, accidents, unstable political situations, and opportunities for nuclear materials to escape control.

But managing the danger better is not an adequate ultimate destination.

Humanity should seek the greatest achievable reduction of the nuclear danger—including, if the necessary conditions can be created, the elimination of nuclear arsenals themselves.

The qualification matters because the nuclear genie cannot be put completely back into the bottle. Even if every existing weapon were dismantled, humanity would retain the knowledge required to build them again. And a world with no nuclear arsenals could become dangerously unstable if one nation could secretly rearm and thereby acquire an overwhelming advantage.

So I would not insist in advance that zero nuclear weapons must be the final destination. It may prove that a world of very small and tightly controlled deterrent forces is safer than a world nominally at zero but vulnerable to secret rearmament. That judgment should depend upon the institutions and verification capabilities humanity succeeds in creating.

What we can say now is that humanity should build the conditions under which nuclear arsenals can safely become progressively less necessary.

Those conditions would include stronger international law, more reliable ways of resolving disputes, verification systems capable of detecting cheating, relationships in which major powers no longer regard one another as plausible existential enemies, and international institutions strong enough that reducing a nuclear deterrent does not invite conquest or coercion.

So the path should be gradual and reciprocal: reduce arsenals where security permits; reduce the danger of rapid launch; strengthen verification; prevent further proliferation; restore and extend arms control; build habits of cooperation; and strengthen lawful means of resolving conflicts. Each successful stage should help create the trust and security necessary for the next.

My central counsel would therefore be:

Do not make immediate abolition the test of moral seriousness, and do not make permanent reliance on nuclear deterrence the test of realism. Preserve enough deterrence to prevent vulnerability now, while deliberately constructing a world in which less and less deterrence is required.

The tragedy of the present nuclear order is that every nation can behave rationally from its own narrow perspective and still help create an irrational condition for humanity as a whole.

An Oracle should keep both truths in view: nations have legitimate security needs, and humankind has a legitimate need not to live forever one mistake, one escalation, or one act of madness away from catastrophe.