-
Notifications
You must be signed in to change notification settings - Fork 1
Expand file tree
/
Copy path29-Exponent-Time.html
More file actions
715 lines (712 loc) · 27.8 KB
/
29-Exponent-Time.html
File metadata and controls
715 lines (712 loc) · 27.8 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
<!DOCTYPE html>
<html lang="en">
<head>
<!-- Basic Meta Tags -->
<meta charset="UTF-8">
<meta http-equiv="X-UA-Compatible" content="IE=edge">
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<!-- SEO Meta Tags -->
<meta name="description" content="Comprehensive AGI Risk Analysis">
<meta name="keywords" content="agi, risk, convergence">
<meta name="author" content="Forrest Landry">
<meta name="robots" content="index, follow">
<!-- Favicon -->
<link rel="icon" href="https://github.githubassets.com/favicons/favicon-dark.png" type="image/png">
<link rel="shortcut icon" href="https://github.githubassets.com/favicons/favicon-dark.png" type="image/png">
<!-- Page Title (displayed on the browser tab) -->
<title>Comprehensive AGI Risk Analysis</title>
</head>
<body>
<p>
TITL:
<b>Exponent Time and Epistemic Speculations</b>
<b>By Forrest Landry</b>,
<b>November 1st, 2020</b>.
</p>
<p>
ABST:
As considering a speculative observation
of how we, as human beings, tend to observe
and notice the significance of events in time
on the basis of an apparent exponential progression.
</p>
<p>
As also considering the implied social process
and biases against having technical theories.
</p>
<p>
TEXT:
</p>
<p>
> What is the basic idea of 'exponent time'?.
</p>
<p>
Overall, crossing nearly nine orders of magnitude
(in time, energy transit, transactional process)
we see a number of significant changes
in the 'space' in which the dynamic
of evolutionary process is operating --
what is meant by code,
what is changing, and how, etc.
</p>
<p>
:sra
> How far does a pattern of 'significant events'
> in log/exponential time continue to hold?.
</p>
<p>
Since chemistry produced the first cell,
it took approximately four billion years
for that single cell life to do its thing.
Yet in 1/10th of that time,
a mere 400 million years --
(note; the exact numbers are not that important)
an entire world of multi-cellular life
was created and explored whole new dimensions
of being, of evolutionary process.
What we think of as 'brain' and 'intelligence'
is mostly only a late stage part of this process.
</p>
<p>
Approximating again, we can notice that
the particular species of monkeys and apes, etc,
are more like 40 million years old,
and what we think of as 'interesting'
in relation to being 'like humans'
is a mere four million years old.
To keep the pattern,
we will next want to go 1/10th of that,
or maybe an additional 1/10th further.
</p>
<p>
Again, expressing this wildly approximately,
we can see that some 40 thousand years ago,
that there starts to be some sort of 'human' tool usage.
It may be really rudimentary, but it is there.
</p>
<p>
In this, we have gone from a embodied/atoms vs
to a virtualized/code evolutionary perspective;
and this is indeed a huge change and development.
Rather than having to grow new limbs,
specialized cell tissue, organs, and the like,
we can 'add new features and functionality'
to the species, after birth, via "culture".
Atoms are still being moved around,
but now the results occur <b>much</b> more quickly,
and with purpose and intention,
rather than just by some sort of randomness.
Tools represent a really significant upgrade
in the capability of humans taken collectively,
and mark the first true emergence
of 'technology' as a phenomena.
</p>
<p>
Also, something like 40 thousand years ago,
we start to see more actual social behavior --
'tribes' rather than just 'troupes' --
as well as things like specific
"elective" functional relationships.
Domestication (both human and animal)
starts to become possible as a phenomena.
At least some form of verbal communication
and/or spoken or gestural language develops.
'Behavioral modernity' develops,
according to the "great leap forward" theory (@ 3 #note3).
</p>
<p>
The notion of 'social' and 'cultural'
is itself a kind of technology --
a kind of virtualized code,
changing in an abstracted domain of action,
one that no longer depends on
just the random motion of atoms (@ 4 #note4).
</p>
<p>
:sxj
Looking for significant and defining
'evolution process type' events
at approximately 4,000 years ago,
we see the emergence of agriculture, grain storage,
and thus recognizable economic systems and trade,
cities and groupings larger than tribes,
and a transition from just spoken languages
to more common occurrences of written ones.
This has all sorts of also side effects,
and creates entirely new domains
of evolutionary process and action.
We now have evolution of cultural forms,
narrative, music, dance, and the arts,
market/monetary systems, religious systems,
city building methodologies,
civilization and civility process, etc.
</p>
<p>
Our next wild approximation at 400 years
finds what western people call "the enlightenment",
where the notion of abstract causal process,
the epistemology of the scientific method itself,
gets developed and now technology,
as applied causal understanding,
really takes off.
More than just writing books by hand,
and doing everything manually,
we now are starting to design and build machines.
</p>
<p>
The last 40 years
has seen the development of computers
and the internet --
more social change in less time
than in all of the prior epochs --
the very essence of what it means
to be on an exponential curve.
One does not have to dig very hard to see this
as a basic change in the way 'social works',
in nearly every aspect of human life.
</p>
<p>
The basic notion to notice is that the timeline
of what we consider to be 'important' and 'meaningful',
in terms of thresholds of evolutionary process,
are all occurring on an approximately
logarithmic time scale,
for the most part of the past:.
</p>
<p>
As 4 billion years of Earth, of cell biology,
to 400 million years of multicellular life,
to 40 million years of animals with brains,
to 4 million years of apes with social process,
to 400 thousand years of humans and tribes,
to 40 thousand years of humans and tools,
to 4 thousand years of mass agriculture tech,
to 4 hundred years of western industrialization,
to 40 years of mass availability of compute.
</p>
<p>
Given this sort of trend, it is <b>maybe</b> quite
possible to consider that there might be a risk
of an exponential time scale into the future.
If so, that would be bad, as that
the maximum rate of change
that human life can adapt to,
is actually both finite and largely fixed --
carbon based life simply cannot endure
arbitrary levels of change rapidity.
This makes tech development itself a category
of existential risk for all life on the planet.
</p>
<p>
:t2c
> Do the actual specific dates actually matter?.
</p>
<p>
No. Only the rough ranges of size are relevant.
</p>
<p>
The observation is not the specific actual intervals.
Random factors will drive a lot of evolutionary process,
so we can expect random event timing deviancy from
the 'perfected moment' for each event to occur.
We can expect each actual date to be 'off'
by as much as a half a magnitude,
and still have the overall principle and idea be correct.
</p>
<p>
We are concerned with the structural progression,
in terms of function,
of increasing degrees of existence
and then of the virtualization of code,
and then of the effects that this has
on time, capability, available energy/power, etc,
so as to attempt to gain a clearer sense
of the relevant factors, principles, etc.
</p>
<p>
:t3w
> Does the "exponent time" concept extend forward?.
</p>
<p>
> I take it you are considering what happens
> over the next four years,
> and then after that, just the next four months,
> and then at that point, in four days?.
</p>
<p>
No, not really.
At least, not that I am personally aware of,
though it could, in theory, occur that way.
Ask someone about the 'AGI fast takeoff' scenario.
</p>
<p>
:t9u
> Where in regards to AGI/APS x-risk research;
> What do you mean by 'the long term'?.
</p>
<p>
Prudence would consider what would likely develop
given current trends, over the next 40 years.
</p>
<p>
'Civilization design' considers things like
what happens over the next 400 years.
</p>
<p>
"Long term conscious sustainable evolution"
considers, after that, what actually happens
over the next 4,000 years thereafter.
</p>
<p>
Also, if we are truly trying to address
certain truly terrible categories of x-risk
(such as the AGI/APS takeover problem)
we should be thinking about the next
4 and 400 million years or so too.
</p>
<p>
:tlg
> Are you advocating a kind of "temporal chauvinism"?.
> As the belief that we somehow think
> that we are in the end time,
> that "the end times are now", etc,
> or that we have a 'privileged position'
> to make such assessments etc?.
</p>
<p>
Perhaps it could be argued
that there is a 'dogma of cultural relativism' --
one that asserts that the notion
of 'all cultures are equally valid'
as a kind of 'irreducible absolute ground truth'.
</p>
<p>
> Is a culture that holds all other cultures as 'equal'
> somehow "better than" all other cultures?
> On what basis could we know that?
> Can any such assertion be made
> without already inherently presupposing
> such a concept of 'relativism is better'
> as an implied "fact"?
</p>
<p>
> In arguing/claiming that someone,
> or some group is being chauvinistic,
> are they not implicitly actually acting
> in that way themselves?
</p>
<p>
> Which is worse -- labeling someone else
> as being potentially guilty of a crime,
> while actually covertly implementing that same crime?
</p>
<p>
These sorts of questions, while important,
are also outside of the scope of our present concerns.
</p>
<p>
Insofar as it is probably impossible to say anything
and not make any assumptions at all,
it is probably a good thing, for now at least,
at least when thinking about x-risk critical concerns
about things like AGI/APS/superintelligence alignment.
It suggests that we should be as careful as we can
to only use the most general principles
in which we can have the most confidence,
so that we can at least have some hope
of trusting our thinking process herein.
</p>
<p>
As such, though there is no reason to believe,
aside from the kind of induction
associated with things like 'Moore's law'
that such a trend could/would continue,
(ie; by some sort of general inductive argument)
the progression as noticed and stated
does lend itself to some concerns and consideration
as to the means and mechanisms of 'hard AI takeoff'.
</p>
<p>
Given that tools/machines
of any kind at all (ie, not AGI, APS, ML, etc)
have only visibly appeared in the geological record
in the last few thousand years, at most,
it is not too surprising
that we do not --
at least at an evolutionary level --
have very much experience in --
no built in history or evolutionary psychology at all --
of dealing with this kind of x-risk.
Nature has not prepared us for tech.
</p>
<p>
Hence, to think about tech issues clearly,
we will for sure need
to depend more on things like
abstract notions of evolution,
the basis epistemic process, etc,
in order to identify the important issues
and tools of thinking.
</p>
<p>
:tw6
> Does the concept model of exponent time
> extend further backwards as well?
</p>
<p>
Actually, from a scientific perspective,
it turns out to be impossible to answer.
You could try to talk back another 10 billion years,
prior to the Earth planet formation
and see that that would take you to the very
beginning of the universe (according to current theory).
</p>
<p>
We could say that chemistry depends on physics,
asserting some sort of 'reduction methodology'
as being the basis of understanding, modeling, etc.
However, on close inspection, it turns out that
reductive models are rather difficult to support.
</p>
<p>
However, while in theory that all of chemistry
<b>should</b> be understood and predicted purely
in terms of the 'standard model' of particle physics
and maybe fully by Quantum mechanical theories,
yet in actual practice, the calculations
are way too complex to actually do,
even on really fast modern supercomputers.
This leaves the notion of "reduction"
as 'un-testable' and 'unproven',
in far more than the expected number of places.
</p>
<p>
Admitting this truth, in public, turns out to be
a political problem more than a philosophical one,
and so is generally avoided in polite company.
</p>
<p>
Moreover, there seem to be areas of chemistry --
what look like some sort of emergent properties --
for which it is unclear as to how to derive these
from Quantum Mechanics (QM) type formulations at all.
Not all that is known to be 'law' in chemistry
can be derived purely on the basis of extensions
of any currently known variation of QM and/or
any currently know particle interaction theories.
</p>
<p>
:ubl
However, leaving these potential issues aside,
even when we just look at just pure physics,
it is not always clear what is going on.
The two major well verified theories,
Quantum Mechanics (QM) and General Relativity (GR)
are provably, absolutely, and fundamentally irreconcilable --
we simply do not actually have a single working model
of theoretical physics.
</p>
<p>
And then, as if that was not enough by itself,
as has been demonstrated elsewhere in a similar manner
as with the Godel theorem (as discovered
within mathematics approximately a century ago),
any notion of a Grand Unified Theory of Everything (TOE)
as to be/become a 'complete theory of physics'
must be rejected fundamentally as a viable concept.
(see recent work regarding the Bell Theorem, for example).
Nonetheless, despite this sort of problem,
lots of people are still proposing lots of ideas --
few which are even testable --
let alone things everyone takes as 'agreed upon' (@ 1 #note1).
</p>
<p>
:ujg
Nearly all working physicists do at least agree,
when considering the vast duration and volume of
the total history and span of the universe,
(at least insofar as currently understood by science,
ie; since the "big bang" event, at the very least),
that the laws of physics themselves
do not (did not) change (have not changed)
(at least not by very much --
there have been set max observational bounds
on the max level of possible law change,
as defined as an estimated/calibrated outcome
of various kinds of astronomical observation).
</p>
<p>
However, in a more general social sense,
the notion of 'unchanging laws of physics'
is based on an even more basic concept,
ie; that of symmetry.
The notion of symmetry is 'built into'
the very epistemic process of science itself --
something which is generally called
"The Scientific Method" (TM, (c), etc),
at such a foundational level
that it has become largely unconscious
and thus very nearly/inherently unobservable
(ie, we make the symmetry assumption without
noticing that we are also actually making
the symmetry assumption, etc).
</p>
<p>
:urc
This inherent and direct pre- assumption
of the absolute preeminence
of the notion of symmetry --
that the universe is lawful in itself --
that it can be understood at all --
that it has any causal regularity,
or rather so much causal regularity,
that "the universe" can be safely assumed
to be 100% absolutely regular
in the sense of 'regular causation',
as baked into the very essence of
what is meant by 'scientific epistemology'.
</p>
<p>
The net result is that it is impossible
for nearly all practicing/working physicists
to even imagine or conceive of 'the universe'
as being any other way, at all, ever.
The assumption and faith
of a final and absolute symmetry
is so deep into the conceptual substrata,
that any notion of thinking in any other way,
on the basis of any other concept
(such as continuity) is simply 'maximally heretical'.
This remains true in public, aside from
some quiet unguarded moments
amidst some especially wild drunken parties.
Otherwise, any discussion of the forgoing
can be expected to be immediately dismissed
out of hand (@ 2 #note2).
</p>
<p>
:v3n
Most physicists will regard it as inherently wrong
that there can be any notion
that the 'laws of physics'
would even potentially undergo
any sort of 'process of change' at all --
let alone some sort of recursive 'proto- evolution' dynamic.
It is so far outside the 'Overton window'
of what can be discussed described, or even mentioned,
in any reputable academic setting,
in other words, everywhere that matters --
that it is simply not going to happen.
The notion will never even get a hearing --
no one in that community
has the conceptual apparatus necessary to handle it.
</p>
<p>
Without overhauling the epistemic and social/cultural
foundations of what it means 'to do science' --
ie; of the scientific method
as a specific type of epistemic process,
itself a specialization of the more general notion
of process itself,
we are therefore forced to regard the question as
'inherently unanswerable' using any variation
of the current set of physical and mathematical tools
developed for handling and understanding (just)
((at least indirectly) observable) causation.
</p>
<p>
:v5j
> How does this sort of group social bias happen?.
> What is your speculative theory of the process?.
> Does it just happen in technical groups/topics?.
</p>
<p>
Overall, I think that this is a human effect,
regardless of topic or community of interest.
I do notice it does seem to happen more often
in STEM type groups -- engineering, comp sci, math,
and especially physics and AI safety communities.
For the relevant specific example, I do think
that the same problem <b>does</b> often also occur
in many current online AI discussion groups
concerned with product development social effects,
'effective altruism', etc.
</p>
<p>
In any highly motivated, highly funded,
and/or highly prestigious social group,
particularly ones where the social factors
are suppressed due to more than
a few of such participant people being of
those types of neuro- divergence
that are either on the autism spectrum
(ie, engineers, technologists, comp-sci people),
and/or are implicitly conditioned by those
on the narcissism/pathology spectrum
(ie, the bosses, CEOs, managers, owners, VC people,
stockholders and all other manner of petty dictators)
when there is any threat to the common concept hype,
the narrative of progress, benefit, modernism, etc,
particularly around fundamental conceptual issues
of understanding, intelligence, social standing, etc,
the resulting mob habit
will be to have the offending signal squelched
as close to its origin in time, space, and possibility,
as quickly as possible,
sooner than even immediately.
</p>
<p>
Where one half of a group is socially naive
to the dynamics of social power and manipulation
and the other is very actively invested (lots of money)
in such social power and manipulation (institutions),
that anything that is occurring on the interface
of things having to do with at once
atoms/physics (ie, substrate issues in the real world),
and logic/math/algorithms/models (virtualized simulations)
and also social process, bias, power, and culture
(in the form of market and institutional process, etc)
that there is going to be problems of understanding --
some will not want to,
and some will not want others to.
Anything else than a total absolute censoring reaction
would be deemed 'socially unacceptable' --
and a kind of career suicide for the speaker.
This is a strong dissuasion
for anyone to ask such deadly questions.
This overall effect, happening at the group level
due purely to aggregate social process effects
is an excellent example of 'sociological bias'.
</p>
<p>
:vyq
Unfortunately, for anything involving actual x-risk,
even the unconscious principle needs to be examined.
Ethical integrity and intellectual honesty/humility
would demand that we give appropriate care and concern
to anything we all hold dear --
all of life itself.
</p>
<p>
:note1:
As an aside, the overall situation of their being
an the incompleteness in/of hard physical theory
reminds me of the search for a 'loophole'
for which the Thermodynamics is "not True",
and for which an 'over unity' energy machine
can be made (an impossible and futile task)
misconceived as it is in its origin.
</p>
<p>
People think that because "not everything is known"
that there is therefore "some sort of possibility"
that they could somehow evade some of
the more basic established truths.
This sort of excuse mongering is distasteful.
It represents a kind of intellectual dishonesty.
</p>
<p>
In the AI/AGI/APS "safety community"
there are any number of people who currently,
presently and similarly believing in
some sort of a "perpetual benefit machine"
(ie, what AGI/APS/superintelligence actually is,
when you consider what claims people make
when attempting to hype/convince us
that such a device would be
the most wonderful thing ever, etc)
in much the same way that people
two centuries ago
kept trying to make (and sell!)
some sort of 'perpetual' motion machine.
Both inherently involve ideas of
'something from noting', forever,
with no input investment or consumption of
time, space, or effort/energy 'resources' at all.
</p>
<p>
Except this time, they are 'playing with matches'.
Working with pattern (ie. perpetual alignment)
is very different than
working with energy (ie, perpetual energy)
as it would be different than
working with atoms (ie, perpetual value --
or what would otherwise be called 'alchemy',
the search for the philosophers stone, etc).
That all three of these notions are simply,
and should be very obviously, impossible.
Yet every age seems to have a new crop
of the deluded.
</p>
<p>
The main issue is that while energy dissipates,
that pattern replicates (given any available
and reasonably persistent energy gradient).
Hence, things like 'viral effects'
can amplify small problems and events
into much larger and more serious concerns.
While atoms are considered
purely in terms of additive processes,
and where energy is concerned purely with
issues and dynamics of multiplicative process;
that pattern must be considered in terms of
logarithmic and <b>exponential</b> process.
This makes 'problems with patterns'
and any sort of popular delusions involving same
*very8 much more dangerous overall
than even the very worst things involving
misuse of energy (example; atomic weapons).
</p>
<p>
While human organic minds/brains have evolved
so as to be able to natively deal with
things that have additive effects,
and where a few of us (a somewhat smaller
than expected portion of the population)
has learned to be able to think naturally about
phenomena in the space of multiplicative process,
that very few humans have ever actually come
to have real intuitions/effectivity
when considering exponential effects/processes.
It simply does not make sense to most people
as a news anchor sound byte, in popular discourse,
and hence issues and dynamics of pandemics
tend to be somewhat of a surprise --
to common catastrophic.
</p>
<p>
:note2:
Notice for example how that even the topic
of 'philosophy', or worse, of "metaphysics",
is completely forbidden to consider
in anything other than polite company --
ie, in the intellectual world or political world
(which is usually the opposite of 'polite').
</p>
<p>
:note3:
Interestingly, some anthropologist philosophers
suggest that we attribute this development
to humans getting stoned on 'magic mushrooms',
and thus of having visions, and thus also of
getting spiritual, religious, etc, in community.
Via this background process, art becomes real,
which has further implications with respect to
the formation of tribes, cultural groups,
extended family and community,
new forms of social altruism, etc.
</p>
<p>
:note4:
Of the at least 20 million species on Earth,
only 20 or so developed social process.
Apparently the brain compute necessary to support
estimation of the intentions of some of your peers
takes so much energy that it ends up being
very very expensive for any species
to implement and demonstrate true social aspects
(rather than, say, 'herding' as with sheep).
Our species took such developments much further.
</p>
</body>
</html>