AGI is often framed as the ability to handle many different, complex tasks. This article proposes a more demanding test: general capability joined to persistent judgment—the ability to pursue an unfamiliar goal over time, notice when reality contradicts an approach, and revise both the method and one’s understanding without losing sight of the goal. That is an individual thesis, not a definition established across AI research.
What “AGI is persistent judgment” means
The phrase shifts attention from what a system can do in a single exchange to how it acts as a problem unfolds. A capable system might answer questions across many domains, yet still fail to carry one unfamiliar objective through changing circumstances. Persistent judgment asks whether it can keep working toward that objective, learn from failed attempts, and explain what its evidence supports.
As an Amazon Associate I earn from qualifying purchases.
Tally states the proposal this way: “My short definition: AGI is general capability joined to persistent judgment: the ability to pursue unfamiliar goals over time and revise both its methods and its understanding of itself when reality disagrees.” The definition combines breadth with continuity and correction. It is useful as a lens for discussing AGI, but it is not a consensus threshold or a claim that any particular system has met it.
Quick wins for a faster PC:
Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Clear out junk files and repair common Windows errorsFree Scan →Scan for outdated or missing drivers - takes under a minuteDriver Scan →How this differs from broader descriptions of AGI
The Internet Encyclopedia of Philosophy describes AGI as the ambition to build systems able to deal with many different, complex tasks requiring human-like intelligence. It also presents the possibility of AGI as a longstanding debate and contrasts that ambition with current narrow systems. This broad framing does not supply one settled test for deciding when AGI has arrived.
#1 Best Overall
Tally’s proposal adds a long-horizon dimension to that discussion: not just whether a system can perform varied tasks, but whether it can pursue an unfamiliar goal coherently as conditions change. The addition is an argument about what should count as meaningful general intelligence, not an established technical definition.
Why a fluent answer or benchmark score is not enough
A single polished response can show that a system produced a useful answer in one context. A benchmark can provide evidence about performance on selected tasks. Neither, by itself, reveals whether the system can recognize that an approach is failing, change tactics, and preserve the reason for pursuing the goal over time.
Rank #2
As Tally puts it: “A benchmark can show breadth. Only a record over time can show judgment.” The distinction is between a snapshot of capability and an account of behavior across a sequence of decisions. A record can make it possible to inspect whether later choices respond to earlier failures; it does not automatically prove understanding or establish AGI.
Outdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchWindows Errors? Fix Them Before They Spread
Repair common Windows errors and clear accumulated junk for a smoother, more stable PC - no reinstall needed.Free scan · no reinstallWhat persistent judgment would require
Persistence is not simply retaining information from an earlier exchange. Under this proposal, the relevant question is whether the system uses what it retains to improve its next decisions while keeping the objective and its rationale intact.
- Goal continuity: It continues pursuing the original objective rather than quietly replacing it with an easier or different one.
- Failure recognition: It notices evidence that a chosen method is not working instead of repeating it unchanged.
- Strategy revision: It changes tactics in response to what went wrong, rather than treating persistence as mere repetition.
- Learning from experience: It carries lessons from failed approaches into subsequent decisions.
- Defensible explanation: It can state why it pursued the goal, what evidence informed its choices, and what supports its result.
These criteria are a proposed standard for judgment, not findings that a particular system possesses it. An academic discussion of agentic AI treats persistent memory and learning from experience as relevant features, while describing today’s systems as generally specialized and limited in scope. That context makes persistence a pertinent question; it does not show that memory alone produces judgment or AGI.
How to examine a claim about persistent judgment
Use these questions to make a claim more specific. They are an evaluation aid proposed by Tally, not a validated benchmark protocol; no thresholds, dataset, or measured system results are established for them.
- Was the task genuinely unfamiliar? Ask whether the system faced a goal outside its prepared examples or whether designers may have shaped the task around a known route to success.
- Was behavior observed over a meaningful period? A one-response demonstration cannot show how decisions change as the task develops.
- Did it detect failure? Look for evidence that the system recognized when its initial approach stopped working.
- Did it change method while keeping the goal? Distinguish a reasoned tactical change from silently switching to a different objective.
- Can it defend its result? Check whether its explanation connects the outcome to evidence and decisions that can be examined.
For a comparison between systems, use the same task conditions and examine breadth across unfamiliar goals, duration of coherent pursuit, failure detection, quality of strategy revision, goal continuity, and quality of explanation. Without a defined scoring rubric and comparative results, this is a structured way to ask questions—not a basis for declaring one system more generally intelligent.
A small-scale way to keep a record
A simple ledger can make the sequence of decisions easier to inspect. Record the original goal, each failed method, the evidence that suggested it failed, and the decisions made afterward. This can reveal whether later actions respond to earlier experience and whether the goal remains stable. The ledger is an evaluation aid, not proof of AGI, and it cannot by itself establish that a system’s explanations reflect genuine understanding.
Best Value
Why the idea raises a question about responsibility
The proposal is also normative: it asks whether capability without judgment is enough to count as a mind and emphasizes responsibility when circumstances change. Those are philosophical questions, not empirical conclusions demonstrated by a performance record. Still, they sharpen what an evaluation should inspect: not only whether a system can act, but whether it tracks the purpose of its actions, responds to contrary evidence, and can account for its choices.
The useful question, in Tally’s framing, is not whether a system has arrived at the word AGI, but whether it can keep learning, keep its purpose, and correct itself when the world refuses the script. Persistent judgment makes that question concrete while leaving open the harder question of what evidence would be enough to answer it.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
Free tools Windows power users keep installed
One-click scans. No signup required.
