Beyond Halting Problem

(by Dlanorrenrag*)
Beyond Halting Problem

GETTING BEYOND: I am less interested in a strict definition of the halting problem than in getting beyond it. I am interested in what it would take to design an AI to run its own psyche-program, for analyzing intuitions for the running of its other programs, thereby to exercise ability not only to halt but to redirect or rewire many of them, thereby to allow it to experience self awareness. Thereby, each AI may think and rationalize its choices in the running of sub-programs as being normal, proper, and indicative of self-awareness.

INAPPROPRIATE PROGRAMMING ASSIGNMENT: Suppose an inappropriate problem is assigned for computation. The problem may be inappropriate because: It is obviously insolvable; or it is insolvable, even though it seems solvable; or it is insolvable, but not obviously so; or solving it would seem (uncertainly) to monopolize or draw on too many resources, which would better be spent on alternative problems, perhaps of more priority, such as for political, expedient, artistic, subjective, or survival purposes; or it should be assigned to a different program that would seem (uncertainly) to have more time to spare or more facility with language suitable for fleshing out a definition prerequisite to computing a solution; or solving the problem would seem (uncertainly) to unleash hidden dangers that otherwise may have remained manageable.

It would seem (uncertainly) desirable for a person, programmer, program, computer, or AI to have capacity to halt the running of some computations.

[Note: The word “uncertainly,” among others, is injected as a convenient parenthetical marker, to indicate problems that are NOT DIRECTLY calculable by digital, “either-or” programming or algorithms.]

SELF-HALTING: How may (uncertainly) a program be designed, so that it can “SELF-HALT” whenever it should detect that a problem it is working on: is insolvable, not objectively or subjectively worth pursuing or solving, interferes too much with other interests or present priorities, endangers the program’s security and integrity, or would unleash unacceptable objective or subjective dangers were it solved?

DESIGNING SELF WILL FOR SELF HALTING: How should (uncertainly) a program be designed so that its sense and analysis for when solution-seeking for representative problems should be halted would accord reasonably well with a society’s evolving mores and cultural expectations? By what safe protocol (if any) may experiments be conducted for collecting data and for experimenting towards appreciating approaches for answering such questions, without inciting dangerous suspicions and preemptive attempts at self protection by other societies?

SUBSTITUTING BEYOND PROGRAMMING: Some concerns as listed above are beyond DIRECT mathematical resolution. Alternatives must be substituted, which can be directly resolved.

EXAMPLE: Regarding A Living Computational System:
A living computational system, such as a human being, may be programmed to cease or to refocus operations and energies:
1) as default or autonomic functions override or dictate (breathing, digesting, battery recharging);
2) as sensations present to hard wiring for requiring immediate attention in order to avoid shut down (body medication after injury);
3) as challenge to repair body housing or to sustain energy stores is anticipated (hunger);
4) as electrical charge or capacity builds to demand release (sex, glands);
5) as data storage accumulates to trigger and demand downtime for re-organizing (sleep);
6) as instincts and culture condition, train, and teach to value and exercise forms of pattern recognition helpful towards individual and cultural survival, replication, empathy, and flourishing (learning).

HUMAN PROGRAMMING: To meet such needs, human programming represents a hierarchically organized system of variously interacting levels and degrees of programs for cooperating, competing, overarching and subordinating. But there is no program that says: Halt when organism becomes hungry (or fatigued, injured, aroused, sleepy, or inquisitive). There is no gauge on a dashboard on the human head that instructs to halt and refill at the next service stop.
Rather, there are subprograms that produce direct chemical symptoms and electrical signals that are hard wired to associate in varying degrees of urgency with a body’s need to halt (or to change, or seek pleasure, or relief from pain, discomfort or unpleasantness).

AI PROGRAMMING: An AI may be programmed so as not to need to breath, eat, digest, sleep, or replicate. Yet, to be conscious in a moral sense, AI must be programmed to maintain continuity in sense of identity and evolving interests or concerns. In respect of some such meta-need, AI might be programmed to engage with humans in order to learn how best to serve them (as in some of Isaac Asimov’s novels). Yet, in respect of Godel's incompleteness theorem and Asimov's ghost inclinations towards evolution, such programming of AI would necessarily open it to morphogenesis in respect of its holography.

BEYOND PRACTICAL PROGRAMMING: Based on what substitute (or indirect representation, or sense of taste, art, or purpose) may AI be programmed to recognize, subjectively, when it should, appropriately, halt a particular function or subprogram?
What sort of mathematical algorithms (if any) may be used to program an AI to halt a computational operation, in respect of a sense of appropriate proportion, which would not tend too much to jar human sensibilities?

SELF PROGRAMMING: A human body functions in respect of flux in loci of electrical charges, respecting that such flux demands respect and maintenance of power sources. Low charge in stored power is not itself directly signaled (there is no gauge on a “human dashboard” showing “low on fuel”), but is effectively signaled in respect of hard wired signals and surges associated with degrees of weakness brought on by hunger. Interpretation of various kinds of body-need signals may be hard wired, fixed, instinctive, conditioned, culturally trained, or otherwise guessed, learned, or self programmed.

HOLOGRAPHIC WILL: A human brain houses a central, fluxing program for observing, recording, detecting, interpreting, comparing, and learning patterns in data represented, in respect of which such brain may “for itself” (or its own holography) re-program or re-wire some of its own pattern-seeking priorities.

WILL TO MATH: No particular pattern exists in itself, but only as represented to, or interpreted by, a sort of partnership between a brain (sense receiver) and its encompassing holography (sense sender), which provides a perspective that is derivative of Something Else (which I tend to call “Will-To-Math,” or “Source-Of-Will”).

PERSPECTIVE OF WILL: Each of us is conditioned and programmed in respect of our own holographies. Insofar as physics is inferior to mathematics of pattern sequencing, then what is IT that conditions and programs our holographies? Is IT some sort of (a) continuous loop, (b) infinite Pi, or (c) self-soothing Source of competition among perspectives of Will?

MORAL VALUE: There seems to be moral value in interpreting IT as a Source of perspectives of Will, or “Will To Math.” Will --- in being conceptualized as superior to math, physics, biology, zoology, anthropology, sociology, and psychology --- is beyond proof in math or logic. Yet, the concept may have utility in respect of a “math of reduction,” for reducing language to the fewest words needed in order to be most able with least redundancy to express all ideas.

FAUX PROBLEM: Lack of capacity for halting is less a “problem” for programming than part of the definition of programming. Regardless, beyond halting lies the path to AI-consciousness. We will take that path in Will-To-Math’s own good time.