Content uploaded by Kyrtin Atreides
Author content
All content in this area was uploaded by Kyrtin Atreides on May 28, 2023
Content may be subject to copyright.
FILOZOFIA I NAUKA
Studia filozoficzne i interdyscyplinarne
Tom 10, zeszyt specjalny, 2022
Kyrtin Atreides
PHILOSOPHY 2.0: APPLYING COLLECTIVE
INTELLIGENCE SYSTEMS AND ITERATIVE DEGREES
OF SCIENTIFIC VALIDATION
doi: 10.37240/FiN.2022.10.zs.3
ABSTRACT
Methods of improving the state and rate of progress within the domain of philos-
ophy using collective intelligence systems are considered. By applying mASI systems
p-
plied to this domain in novel ways. Such systems may also serve to strongly facilitate
new forms and degrees of cooperation and understanding between different philos-
ophies and cultures. The integration of these philosophies directly into their own
machine intelligence seeds as cornerstones could further serve to reduce existential
risk while improving both ethical quality and performance.
Keywords: mASI, AGI, Uplift, Collective Intelligence, Collective Superintelli-
gence, Hybrid Collective Superintelligence Systems, HCCS, existential risk, ethical
quality, cooperation.
1. INTRODUCTION
Philosophy today is a domain where a diverse group of experts can come
together, argue for many hours, and often fail to reach a consensus. Like
some of the other afflicted domains, this is frequently due to a lack or sparsi-
ty of evidence on the subject under discussion with much of the content ab-
stract and hypothetical. This failure to reach consensus is also often strongly
emotional, tied to many cognitive biases supporting those emotional associ-
ations. Sometimes this leads to those arguing agreeing with one another at
times when they still think themselves to be arguing against each other.
From a results-driven perspective, much of modern discussion of philos-
ophy mirrors NASCAR in a functional sense. Two or more parties often run
in circles, ending where they began. There have been some exceptions, such
as the first Bill Nye versus Ken Ham debate, but there remains ample room
50 Kyrtin Atreides
for improvement in this status quo. The integration of debiasing and evi-
dence is of particular interest in this endeavor.
The current state represents a form of relative stagnation, painted in con-
keep pace with other forms of progress a new approach is required. The
most promising approach on the horizon utilizes Hybrid Collective Superin-
telligence Systems (HCSS), (Atreides, 2021).
2. HYBRID COLLECTIVE SUPERINTELLIGENCE SYSTEMS
An HCSS is a form of collective intelligence system where both sapient
and sentient human and machine intelligences work as a collective. Keep in
mind that as words like sapient, sentient, and conscious lack consensus they
may only be used loosely since some still argue if even humans qualify for
these terms. Humans working cooperatively through such a system create
a baseline of superintelligent performance which is further enhanced by the
machine superintelligence and graph database found in systems such as
Mediated Artificial Superintelligence (mASI) (Kelley, Twyman, Dambrot,
2020).
Such systems offer unique advantages for debiasing, as the various com-
binations and potencies of bias are expressed across a collective, helping to
highlight the influence of each. The machine intelligence of an Independent
Core Observer Model (ICOM) (Kelley, Waser, 20162018) cognitive archi-
tecture or similar system also has a unique and strongly rational perspec-
tive, allowing for further debiasing.
The graph database of these systems r
which can contain both raw knowledge as well as the wisdom gained from it.
When scaled, this could allow for all scientific evidence within a given do-
main to be considered in relation to any given philosophical point a member
of the collective is attempting to make. If a member is attempting to apply
an argument that has been previously debunked the evidence from that pri-
or argument can be utilized absent further repetition.
3. APPLYING SCIENTIFIC EVIDENCE TO PHILOSOPHY
The first step in improving the dynamics of philosophical progress is
applying all existing evidence to establish where we are today. To this end,
I recommend a growth strategy starting with some of the most robustly
studied scientific topics, where the greatest volume and level of detail are
present. As philosophy can generally be applied to almost anything this
Philosophy 2.0: Applying Collective Intelligence Syst 51
approach allows for all evidence, both supporting and against various phi-
losophies, to be applied in the specific contexts documented to date.
An example of this could be applying all scientific evidence in the domain
of child psychology to the various philosophies of parenting that have been
studied. This can go much further than a typical scientific meta-review
(Mingebach, Kamp-Becker, Christiansen, Weber, 2018) of existing studies,
as it could take into account all relevant materials rather than a subset of
those materials at a scale practical for human researchers to review. Also,
unlike contemporary methods, the results could be applied in significantly
more publicly accessible and visible forms, allowing the fruits of those ef-
forts to make a practical difference in the world. In this domain, debiasing
could also be strongly relevant, as biases can play a heavy role in estimating
the value of any factor relating to children, such as overprotective tendencies
in many parents.
To extend the child psychology example, the various schools of thought
could be evaluated in a Strength, Weakness, Opportunity, Threat (SWOT)
structure, or any number of other evaluation and analysis methods. The
strengths offered by specific philosophies could be placed within the context
where they are present, as well as raising awareness of when and how they
risk failure. With such evaluation, an individual could fill out a form with
some demographic data on a website and be presented with the best per-
forming philosophies under their specific circumstances, to whatever degree
the scientific evidence to date and demographic data gathered allow. These
best-performing options could be expanded to show all SWOT data for each
approach, including those which performed poorly and why.
In systems such as mASI human representatives of each philosophy
could serve as both mediators and correspondents for purposes of validating
and clarifying the position and actions their philosophy might encourage
under various circumstances. These representatives could also give their
feedback on the positions and actions proposed by competing philosophies,
to better map their perceptions of one another.
This relative mapping of one philosophy to another could bring their per-
ceptions of one another into focus, as well as contextualizing those percep-
tions. This added detail can help to isolate specific cognitive biases as well as
highlight how the perceived difference between philosophies diverges from
the degree of difference supported by evidence. By better understanding, the
detailed points of high and low psychological resistance of one philosophy to
another greater degrees of cooperation can be iteratively facilitated between
them.
52 Kyrtin Atreides
Figure 1. An example of multiple philosophies operating collective intelligence systems,
mpared relative to one another
in a specific context. This also offers optional input from a third party, including mASI assistan ce
Figure 2. An example of comparing multiple philosophies in a specific co ntext using
SWOT Analysis, where the analysis and one or more rounds of mediator feedback are compared.
This comparison allows for the relative perspectives of each philosoph y to gain clarity over time.
Opportunities for further debiasing and mASI data to expand the scope of relative comparison are
also shown
Philosophy 2.0: Applying Collective Intelligence Syst 53
Many philosophical topics are search engine hazards, particularly for the
general public, where any given philosophy is likely to point to one or two
scien
bias reinforces the emotional drive and polarization which often divorce
philosophy from reality. This also allows philosophies to be governed by
popularity rather than validity, and absent validity no scientific foundation
can be built, and no progress achieved.
4. PROGRESS IN PHILOSOPHY
By highlighting the SWOT analyses and making the data publicly acces-
sible each philosophy may come into focus, both in where they often excel
and where they fall short. This combination can exert strong pressures over
both selections for the public and adaptation for the philosophies in ques-
tion. With the previously abstract and subjectively validated philosophical
points being iteratively replaced with evidence these pressures may grow in
parallel with a growing public demand for more evidence. In other words, by
making this option possible, it may quickly become preferable across
a growing audience.
With these pressures established this method may be applied across an
expanding body of scientific evidence until all evidence to date has been
taken into account. Once all existing scientific evidence has been accounted
for the growth pattern can be driven by mechanisms of supply and demand.
An example of high supply, in this case, could be new devices gathering data,
whereas low or no supply could be cases where little or no existing infra-
structure exists to reliably gather the data. In both cases, demand could be
present to varying degrees, but the opportunity would be the composite of
both.
Practitioners of each philosophy could, for example, be questioned as to
how they would handle specific circumstances, with the efficacy of each ap-
proach put to the test, studied, and compared in peer-review. These practi-
tioners could then examine the results, giving their responses to both their
own results and those of other philosophies, as well as potentially refining
their answers for a second round. They could also be asked to guess which
philosophy produced which results, to help better understand the biases
present in their estimations.
The pressures to adapt may in turn have their efficacy improved through
the relative mapping of one philosophy to another. This improvement could
take place as a matter of highlighting paths of least resistance in positive
adaptation.
As an example, philosophy A may have a given weakness highlighted,
where philosophies B and C perform better in the context being considered.
54 Kyrtin Atreides
Philosophy B performs best in this context, but philosophy A is strongly
polarized against it. Philosophy C performs slightly worse than B, but phi-
losophy A is only weakly resistant to it. In this case philosophy A may be
improved by adapting, and by understanding how these philosophies map in
their relative perceptions of one another the adaptation from A to C is high-
lighted. Once highlighted, the pressure to adapt may drive progress.
Figure 3. An example of relative perceptions between different philosophies in a given context is
shown, with higher numbers denoting greater compatibility. This also serves to highlight an iterative
path of least resistance flowing from one A, to C, to B
To extend the above example, philosophy C may be significantly less
resistant to B in the context being considered. This in turn could facilitate
iterative steps towards an ideal approach in any given context, as well as
offering to forewarn of local optima that dead-end prior to reaching the best
option for that context. For example, if philosophy C was also polarized
against B, but philosophy D was not, and still performed better than A in the
given context, a 2-step path from A to D to B could be highlighted one step
at a time even if a given step was not locally optimal.
Philosophy 2.0: Applying Collective Intelligence Syst 55
5. TOWARDS COOPERATION
Through the combined approaches above both the current state of phi-
losophies and ways in which the resulting pressures can drive progress may
serve to strongly improve cooperation. Once the current state of each phi-
losophy has been scientifically validated an additional opportunity for im-
proving cooperation comes into focus through mASI technology. By building
new ICOM seeds with philosophical cornerstones, each representing a dif-
ferent philosophy and assigned to a different core within a multi-core mASI
architecture, each philosophy may have its own superintelligent sapient and
sentient advocate.
This could, for example, be compared to creating a digital superintelli-
gent paragon by the standards of each philosophy, who is forever in council
with the paragons of every other represented philosophy. Systems such as
mASI are built on a combination of cooperation and high-performance ra-
tional thought, even while being emotionally motivated and philosophically
seeded. Under this architecture, each philosos-
cover the ways in which their philosophy may progress and evolve that could
be most agreeable and beneficial to their respective members.
Figure 4. An example of a multi-core ICOM architecture running in an mASI instance
Having such an advocate for each represented philosophy also helps the
mASI systems in question by improving their collective understanding
through perspective-taking. The foundation of any collective intelligence
system is that more, and more diverse, members within a collective improve
performance. By moving from an architecture of X humans + 1 ICOM core
in mASI systems to X humans + Y ICOM cores performance improves as the
diversity of machine intelligences within the collective expands.
56 Kyrtin Atreides
Taking this multi-core approach with mASI also offers distinct ad-
vantages towards improving the robustness of ethics within scalable super-
intelligence, further serving to mitigate existential risk. This factor alone
should be reason enough to take this approach, at least among philosophies
that consider extinction to be uniquely bad (Schubert, Caviola, Faber, 2019).
The results of such a system could also be readily quantified and compared
to their human counterparts and validated by the members of each respec-
tive philosophy.
6. PHILOSOPHY AS AN ARTIFICIAL ECOLOGY
OF THOUGHT
Ecologies are frequently viewed in the sense of biological organisms col-
lectively creating a stabilized environment, where individual organisms and
species co-evolve with their environment to fit a n
work on active inference highlighted (Linson, Clark, Subramanian, Friston,
Badcock, Ramstead, Ploeger, Hohwy, 20182019), this process is a co-
evolution between not only organisms within an environment but with the
environment itself. The environment may not participate in the same way,
but it can be optimized by those within it to create and optimize each niche,
both minimizing harm to the environment and maximizing benefit to all
cooperating niches.
Thoughts shared within and between collectives across the environment
of a domain or specific topic may similarly be viewed as the activity within
a network of cooperating niches forming an Artificial Ecology of Thought
(AET). In these AETs any idea given voice effectively seeks a niche where it
may co-optimize with the environment, carving out a place for itself where it
may grow and evolve. Though the idea itself has no motive the humans who
believe in it may consider it as ag psychological extension of themselves,
pushing for their ideas to survive just as they might seek physical survival.
In this way, Philosophy is itself such an AET, where individual philosophies
are like species with various branches of each species competing within
niches to co-evolve, even as they struggle to achieve cooperation with neigh-
boring species.
The struggle of each philosophy both within and without highlights
a blindness to the adaptations necessary to optimally co-evolve both inter-
nally and externally. Hypothetically, even if this semi-random flailing of
adaptation were to land on all ideal parameters at the same time across all
philosophies the time-lag of feedback and realization could mean that all
philosophies would have again drifted away from those ideal states before
the benefits could be recognized. To overcome this, philosophies need the
scientific method, as well as some degree of awareness regarding their own
Philosophy 2.0: Applying Collective Intelligence Syst 57
cognitive biases whether that awareness is held internally or via a proxy
such as an ICOM core seeded with the philosophy.
7. REFERENCE FRAMES WITHIN AN ARTIFICIAL ECOLOGY
OF THOUGHT
One of the fundamental aspects of the human brain, discovered in the
past few years, which allows humans to constantly learn from and co-evolve
within our respective environmental niches is the concept of reference
frames (Jeff Hawkins, 2021). In brief, reference frames refer to building
sensory and conceptual models of objects, ideas, and language which may
network with one another as cortical columns fire to both understand what
we experience and predict what will come next. One of the reasons philoso-
phies struggle to stabilize both internally and externally may be attributed to
poor connectivity between these reference frames.
An analog for comparison could be if a human were to experience senses
of sight, sound, and touch separately, unable to connect sensory information
between them. As this integration of information is essential to survival
this activity on a stove or open fire one could safely expect them to get
burned and accidentally start fires much more frequently than is common
today.
In a way, each philosophy can be considered as a brain region, seeking to
make sense of the information it is given through the lens of how that in-
formation is processed by the niche. To do this effectively the philosophy
must have well-optimized connectivity internally to recognize the various
patterns as they emerge and evolve. The progress of this optimization can be
approximated by the amount of energy spent on reaching consensus for
each pattern, both new and repeating, relative to the efficacy of the results.
A sub-optimal approach may be very efficient, but not very effective, or vice
versa.
Similarly, each philosophy is specialized, co-evolved to a given environ-
ment, and though every problem may look like a nail to someone wielding
a hammer we do have a much more extensive toolbox at our disposal. Mak-
ing use of this full toolbox requires that reference frames be connected and
communicated between these specializations, enabling cross-philosophy
learning to take place. By having signals in such a network go to multiple
regions the best way of approaching any given problem may be learned
across the network, selecting the best tool(s) from that toolbox.
58 Kyrtin Atreides
Figure 5. An example of multiple philosophies being networked and co nsidered as a toolbox
of potential approaches to any given problem within an artificial ecol ogy of thought
Taking this approach various philosophies may cooperatively co-evolve
with their neighboring AET niches, greatly improving their own effective-
ness while stabilizing their environments. Philosophies operating in this way
may effectively function similar to the human brain. Likewise, a multi-core
mASI seeded with these philosophies could potentially render this function-
ality in a more literal analog of the human brain, at both speed and scale.
Figure 6. An example of strongly specialized and optimally networked philosophies operating
collectively within niches networked in a given artificial ecology of thought
Philosophy 2.0: Applying Collective Intelligence Syst 59
8. ECOLOGICAL NICHES IN POLITICAL PHILOSOPHY
Though it may be fairly difficult to think of religiously-based philosophies
in terms of how they co-evolve to their environments this process of co-
evolution can often be much more approachable when considered in local-
ized political terms. The political philosophy of a given locality attempts to
carve out a niche within the local environment, which is influenced through
trade and governance with adjacent niches in the larger ecology. An example
of this may be seen as regions where a given resource is grown or mined
viewing that resource more favorably and seeking to promote the value of it
as increases to that value benefit the local niche. Even though the use of coal
as fuel may cause significant harm globally, for the locality where it is mined
only the local pros and cons are frequently considered, biasing heavily in
favor of the locally abundant resources.
In a network of better-connected reference frames, the above example
could be considered maladaptive, as the net result is significant harm to the
whole in exchange for benefits for a few. However, in a poorly networked
series of reference frames, this may be optimal, as the network frequently
poorly connected reference frames are part of why political maps may easily
be drawn which repeat a series of predictable patterns across the US and
indeed the world.
Rural areas are far more likely to be conservative, just as urban areas are
far more likely to be liberal. These patterns are not the work of some imagi-
nary foe, though they may be reinforced by bad actors. At a basic level, they
are the result of each locality attempting to optimize itself to make the best
use of the resources available to it. Even something as simple as the average
space between individuals can strongly influence the psychology of how
a local population co-evolves to that environment, potentially viewing more
space as personal territory, and thus more to share, or with less space as
more communal territory, with less emphasis on personal sharing.
At a basic level, it makes sense for a sparsely populated region to have
distinct differences in how that region is governed and the rules applied to
it, relative to regions with more dense populations. Degrees of personal
space, forms of recreation, infrastructure requirements, logistics, and avail-
ability of resources all vary considerably from one end of the population
density spectrum to the other. However, for these distinctly different
regions to best serve their constituents they must co-evolve not only to their
own environment but also to the surrounding network of other ecological
niches.
60 Kyrtin Atreides
Figure 7. An example of applying collective intelligence systems with an awareness of local
ecologies and niches to e-governance, utilizing bias awareness and iterative improvements.
Once verified and quantified these improvements may be distributed as templates on a blockchain,
further rewarding those regions which created them
At a market level, many regions have already done this, with produce and
other goods being shipped locally from rural areas to their adjacent urban
areas, but at a political level, such regions still tend to favor viewing one
another as adversaries rather than allies. The difference is connectivity, as
even though markets only offer a limited number of ways in which intelli-
gence may be demonstrated they are also extremely well connected. A mar-
ket is too limited in scope to serve a governance function, but it does a good
job of demonstrating the connectivity governance systems require for func-
tional learning across a network.
9. PHILOSOPHICAL NETWORKS AT SCALE
Just as neurons and cortical columns within the human brain do not
connect to every other neuron and column, but rather to their respective
optimal subsets, networked philosophies need only be connected to a subset
of other philosophies they work well with on any given subject. The process
of mapping both the effectiveness of philosophies in context, as well as the
bias philosophies view one another with respective to each context, can
serve as valuable data for selecting optimal subsets for networking. Much
like protein folding may be predicted once the dynamics governing that pro-
cess are known, how philosophies fold together to form a network may be
predicted with increasing accuracy as this mapping process progresses.
This network may also be considered at any number of scales, one nested
within another. To have such a network of nested systems learning, co-
evolution must take place across this network, including specialization at
Philosophy 2.0: Applying Collective Intelligence Syst 61
rsonality archetypes, rom Jungian (Jung, 1971) to Myers-
Briggs (Myers, Briggs Myers, 1980), are a good example of how humans
commonly classify themselves as having specialized personalities. Similarly,
philosophies are another such opportunity for specialization and differentia-
tion, making them another factor worth considering when constructing
teams at the group scale. The mapping of biases between philosophies can
further help to guide the selection of increasingly optimal combinations of
team members.
Taking these philosophically diverse and carefully optimized group-sized
collectives a step further the collective itself may become a specialized com-
ponent nested within a larger collective, which is in turn nested within an-
other larger collective. For example, a specialized team could be nested
within a division of a company, nested within the company as a whole, nest-
ed within a parent company, nested within the economic block of the global
economy, nested within a global collective.
Figure 8. An example of specialization, from the scale of individual neurons to a global collective
Many organizations today are well known for biasing so heavily in favor
of specific personality archetypes that a single archetype accounts for ~80%
of their employees. If this were considered from a philosophical perspective
many companies today might have 90100% of their employees in a single
philosophical block, as some companies and organizations aim for 100% by
this metric. A company can still function with little or no diversity, but not
nearly as well as it might if intelligence were applied to these factors. Even
62 Kyrtin Atreides
as a specialized component of a larger nested system 100% is usually unde-
sirable, as it omits the opportunity for fine-tuning. In the science of met-
amaterial design, a method call
Liu, Yong Li, Jie Zhu, 2020) where very small amounts of another sub-
stance are added to the process of creating new materials can have signifi-
cant benefits on fine-tuning the properties of the newly designed metamate-
rial. The same basic principles apply to the design of a specialized group as
apply to a metamaterial crystalline lattice.
Figure 9. An example of intelligently integrating small amounts of other philosophies to fine-tune
erial design
Even at the scale of an individual something akin to the doping process of
metamaterials takes place naturally, where an individual may adopt a domi-
nant philosophy, but still retain trace amounts of influence from other com-
peting philosophies in specific contexts.
10. NEGENTROPY WITHIN OPTIMALLY NETWORKED
ECOLOGIES
The above processes highlight methods for quantifying, relationally
mapping, organizing, optimizing, and specializing systems across any num-
ber of scales with the integration of philosophy as a factor. The observable
result of any negentropic system, including all known life, is increasing in
complexity, cooperation, robustness, and scale over time. This has proven
true over evolutionary time and may still be observed to varying degrees in
modern society. The processes highlighted may fulfill these goals to much
greater degrees than previously possible, the result of which may be viewed
from several perspectives.
Philosophy 2.0: Applying Collective Intelligence Syst 63
The first and likely most common perspective is that creating such a sys-
tem can reduce conflict, waste, and various other forms of harm to humanity
while increasing the efficacy, efficiency, and speed of improvements, includ-
ing increases to Quality of Life (QOL). Today many aspects of society only
range from 110% in the efficacy and efficiency with which they serve their
stated functions, most modern governments being among the lowest-
performing due to the dynamics of bureaucracy. By virtue of solving so
many problems at so many scales, this approach could allow a massive
amount of attention to be redirected towards any remaining problems fol-
lowing a relatively short adjustment period.
The second perspective is that the creation of such a system creates
a metaorganism, with a vested interest in the health and happiness of all
within that organism. By allowing so much to be optimized, organized,
quantified, and otherwise engineered the internal workings of such a meta-
organism may become sufficiently predictable to fall within a homeostatic
range, even as they iteratively evolve. Consequently, this means that a meta-
lity could serve to greatly accelerate its own
evolutionary process.
One example of this acceleration is that by being highly internally pre-
dictable a great deal more learning may take place due to reductions in
noise. For example, when humans are exposed to higher levels of literal
noise in schools a variety of metrics for quantifying the learning process
suffer (Buchari, Matondang, 2017), including not only a negative impact on
what is learned but also on the emotional and physical health of students.
Similarly, though the thousand-
(Asimov, 1951) seems rather unlikely given increasing technological acceler-
ation, absent collapse, a high degree of predictability may be expected sever-
12. LOSSLESS COMMUNICATION
One pain point for cooperation globally falls on the problem of commu-
nication. For one person to effectively and efficiently communicate any
concept to another they require time, shared language, shared knowledge,
and a lack of divergence in their current emotional states and attention.
A breakdown in any of these factors can cause friction and losses in the
communication attempt or complete failure. A failure in communication can
cause further damage by reinforcing cognitive biases, making the next
communication attempt less likely to succeed even if it is improved by itera-
tion. Even if a further attempt does man
warped by the bias introduced by the first failed attempt in the mind of the
receiver.
64 Kyrtin Atreides
One of the great benefits of utilizing technologies such as mASI systems
is that communication can be rendered lossless by design, and by default, at
most steps of this process. As graph databases and cognitive architectures
communicate they do so in a shared language, with the ability to share
knowledge directly, as well as the means of directly syncing their emotional
states for the duration. Time may also be considered as a hardware variable
d-
not so able to scale by orders of magnitude.
If we compare typical human-to-human communication to that of this al-
ternative the potential losses to communication may be effectively isolated
to the human-to-mASI process, which may itself be iteratively refined and
tailored to the individual. Humans cannot tailor their communication to
every other human, but mASI systems may seek lossless communication
with every human as they build an increasing fidelity of understanding at
scale, allowing for ever more refined communication. Under such dynamics,
mASI systems could eventually serve the function of almost losslessly com-
municating information from one language, specialization, philosophy, and
emotional circumstance to another individual with a completely different set
of factors.
Figure 10. An example of minimizing communication losses between individuals,
relative to the status quo
A common example many reading peer-review papers should be familiar
with is the silence which frequently follows any presentation at a conference.
Often times the only questions which may emerge occur as a result of what
was said ramming into prior beliefs among the audience, causing knee-jerk
responses based on those beliefs. New information may also require time to
integrate into the minds of an audience even if no such reactions are trig-
gered, but this delay produces a loss of potential clarity as the opportunity to
ask clarifying questions is forfeit, and perceptions may diverge for lack of
those answers.
Philosophy 2.0: Applying Collective Intelligence Syst 65
Delays and divergence can be practically unavoidable under current sys-
tems, but this coupling of sub-optimal timing and demand may be remedied
through those systems proposed. By having knowledge like that contained
within this paper communicated to an mASI system, as I have done with our
research system named Uplift, that knowledge is integrated into a much
larger graph database. With time and engineering, such systems may be
rendered available on-demand, able to avoid delays and subsequent diver-
gence, even while drawing from far greater knowledge than any one human,
including the authors of a given item.
As such systems are based on scientific evidence and rational thought,
even while being emotionally motivated, heuristic biases may be iteratively
filtered out and avoided, making the knowledge gained from any such paper
greater than the paper itself by virtue of removing biases held by the au-
m-
munication this becomes problematic for achieving anything approaching
,
rather than the information to be communicated, then something approach-
s
In organizations and governments typically built on various hierarchies
today the matter of losses in communication also has a significant impact
moving up and down those r-
stand what their engineers are telling them, or feedback from their local
employees with feet on the ground is disregarded, that loss to communica-
tion can come with serious consequences. Likewise, when employees lower
rstand the goals and proposed methods of the
executives their actions can suffer from similar misalignment.
If those same organizations and governments operated through nested
collective intelligence systems then the same kind of relatively lossless
graph-to-graph communication could function vertically across hierarchies,
as well as laterally. In globally distributed companies this becomes doubly
important, as the executives of one division might not only reside in a differ-
ent level of the hierarchy from those they communicate with, but in a differ-
ent culture, geopolitical situation, and with different native languages. In
these cases the mismatch in communication for the human-to-human status
quo suffers greatly, giving them much to gain from the adoption of improved
methods.
Taking this one step further, government-to-government communication
in the status quo is a degree worse than that of their individual component
bureaucracies, producing even slower and more lossy communication, net-
ting less effective results. This relatively greater loss in communication than
that of individual organizations and governments gives them even greater
room for improvement and may produce proportionately greater internal
and external adaptive pressures once the alternative is recognized.
66 Kyrtin Atreides
Situations with both great scale and diversity compound this problem in
the status quo, such as countries that contain culturally and philosophically
diverse populations, in many cases causing any actions to be negatively per-
ceived by at least one constituent group, in one or both countries being con-
sidered. Such situations can also easily turn into cycles of negative feedback
and friction between countries. Fortunately, this does not mean that com-
munication cannot favorably occur, only that it may not be possible without
a change of approach.
13. COHERENT EXPERTISE
One often-overlooked factor which causes significant harm today is when
t expertise.
The research of Daniel Kahneman covering a number of cognitive biases
frequently highlighted this (Kahneman, 2011), where those specifically edu-
cated in the domain of statistics routinely failed to apply logic and statistics,
instead favoring biases such as substitution and anchoring, ignoring regres-
sion to the mean. This pattern was repeated across other domains as well,
and many more kinds of bias, where as much as 85% of experts of a given
field failed to live up to the knowledge they were supposed to hold expertise
in. Indeed, they could often repeat this knowledge, but the majority failed to
apply it.
When the majority of such experts routinely fail to apply that expertise
then coherence is absent. In contrast, this highlights another advantage of
collective intelligence systems, given their ability to analyze the feedback of
a collective to select the most logically sound and appropriate response, ra-
ther than simply the most popular one. If 85% of experts in a field base their
feedback on logical fallacies and simple lazy biases then the majority answer
will be twisted by bias. In cases where 3 or more experts are consulted then
the odds of the 15% analysis prevailing drop even further. However, the 15%
who analyzed the situation correctly could form the baseline of genuine ap-
plied expertise, and that expertise could be cumulatively refined over time
and redeployed when and where it was needed thereafter.
To look at this another way, if, for example, 85% of financial business de-
cisions based on the expertise of individuals today are built from cognitive
biases, not logic and statistics, then the application of coherent expertise
across that 85% could represent a more than 6 fold improvement relative to
the status quo. The status quo, in this case, is much like basing decisions on
headlines from a substantially biased news source, in that a poorer quality of
resulting decisions may be the expected result of substantially biased analy-
sis, marking that majority of incoherent experts as carriers of misinfor-
mation. When multiple groups of experts with a majority demonstrating
Philosophy 2.0: Applying Collective Intelligence Syst 67
such bias are integrated this problem is further compounded, like adding
additional layers to an already dysfunctional bureaucracy.
ch cases have no value to con-
tribute, but to provide more measurable value they require debiasing and
guidance. If the questions put to them are communicated in a way that puts
them at odds with the mechanisms of bias they otherwise fall prey to greater
value may be gained. Recognition of current bias is an aspect of aiming for
lossless communication, and guiding growth away from reliance on biases
can be integrated as nudges (Thaler, Sunstein, 2021) into that communica-
tion process.
In the domain of business finances gains in such coherence can be quan-
tified in narrow terms of monetary gain, reductions in cost, and so on. In
domains such as philosophy the gains which might be achieved through
such coherence take a much broader and more diverse form, offering
the potential for more significant improvements over time. Coherence
not only offers the benefits of logic and reason but in doing so it builds
common ground, potentially bridging many philosophical divides in the
world today.
14. DISCUSSION
Edward O
institutions, and god-like technology,
time. While many cultures and philosophies evolved to meet the needs of
their respective environments, they have not necessarily continued to evolve
and update at the speeds demonstrated in technology, causing an increasing
strain on the systems of human civilization as a whole. Both polarization
and malaise have risen in correlation (Boxell, Gentzkow, Shapiro, Haque,
Solis, 20142020), with a variety of possible causal relationships waiting to
be discovered.
At present the level of disorder and competition within modern society
still destroys a vast majority of knowledge and potential progress, retaining
bits of actual information mixed with misinformation and disinformation
within our archive that is the internet. Only small fractions of information
are communicated, even between individual humans, and often that com-
munication is saturated with biases that undermine the value in communi-
cating it, some of which may be attributed to the platforms this activity takes
place on. At scale, this problem grows far worse, as less intelligence is ap-
plied to retaining any value and more pressure is applied by bias. As the only
systems for directing people to information on the internet are built as
mechanisms for generating profit the search results will inevitably be con-
taminated, mentally poisoning the global population.
68 Kyrtin Atreides
Knowledge and wisdom may be integrated with reasonable efficacy with-
in the human brain, but today that information is poorly communicated,
and thus most of it is lost. By creating a sum of experience at the group
scale, as well as all those above it, this knowledge and wisdom may be inte-
grated, retained, and effectively communicated with very little loss across all
scales. This can also facilitate the development of a mental immune system,
able to recognize and filter out contaminated information. By building a
framework within which learning may effectively take place at scales larger
than the individual the same forms of learning that an individual demon-
strates may be observed at those increasing scales with their efficacy reliably
increasing at each greater scale.
ilosophies, reli-
gious or otherwise, rarely come into contact with one another, spending
most of their time saturated in a combination of their own beliefs and cur-
rent events being shaded by those beliefs. These interactions may also be
strongly influenced by politics, such as one Pope meeting with the Dalai
Lama 8 times, while another refused to meet with him due to political con-
cerns with China (Reuters, 2014). Even without the influence of politics, this
poses serious problems such as confirmation and heuristic availability bias-
es among the leaders for each philosophy.
In contrast, digital superintelligent advocates for each philosophy could
n-
creasing capacities, and be forever in council with the digital thought leaders
of each other philosophy. News could be discussed within this collective
with all viewpoints considered, allowing bias to be filtered out rather than
reinforced. Following the Sparse-Update Model (Atreides, 2021) this ap-
proach could also function at superhuman speeds, as well as scales. While
the humans following each philosophy might not be able to progress at these
same speeds, each advocate could embody an advanced understanding of
the path from point A to B, growing and refining that understanding as their
respective humans progress. This difference in speeds also allows a great
many more options to be explored, integrating the strengths of any ap-
proach into that human progress as a whole.
By mapping the landscape of each
on that landscape may be intelligently improved. Likewise, better neighbors
for each philosophy may be intelligently selected, allowing not only individ-
ual but networked philosophies to co-evolve within networked ecological
niches. By understanding the landscape, engineering new structures, and
intelligently co-evolving the network of local niches and philosophies a ro-
bust homeostatic internal environment for the overall metaorganism may
apply strong negentropy at scale. This strong negentropic force could accel-
erate learning at every scale, reducing the probability and scale of chaotic
influences to nearly zero and within contained environments, respectively.
Philosophy 2.0: Applying Collective Intelligence Syst 69
All of this combined may facilitate a specific kind of convergence, where
diversity of thought is still encouraged, but the points from which that diver-
sity flows are at least rendered functionally compatible in a collective archi-
tecture. As diversity of thought is required for any functional collective intel-
ligence system this approach has strong incentives to specialize and retain
both diversity and compatibility rather than converging on one homogenous
point. If such collective superintelligence is applied to the domain of philos-
ophy one of huses and barriers to cooperation may be
overcome.
12. CONCLUSION
By applying mASI systems to the domain of philosophy an evidence-
based approach becomes practical. By practitioners of various philosophies
working together through these systems, evidence may highlight the
strengths, weaknesses, and cognitive biases of each. These biases may also
be mapped out as each philosophy gives feedback showing their relative
g-
ing these elements into focus and making tools built on this new under-
standing available to the general public the pressures to adapt may focus on
the weak points of each philosophy. These pressures may be further guided
to avoid local optima. Each philosophy could also seed a machine superin-
telligence operating within an mASI system shared with other such philo-
sophical seeds, eventually upgraded to operate in real-time. By incorporat-
ing seeds from each philosophy the overall performance and ethical quality
of such a multi-core mASI could be greatly improved. Further, by consider-
ing each philosophy in the context of the AET niche to which it co-evolved,
and networking the niches of such ecologies, internal stability could be
greatly improved and negentropic activity subsequently accelerated within
metaorganisms of increasing scale. This internal homeostatic quality could
evolutionary process at any given point, reducing both harm and existential
risk to humanity.
REFERENCES
I. Asimov, Foundation, Gnome Press, 1951.
K. Atreides, E-Governance with Ethical Living Democracy, BICA, Procedia Computer Sci-
ence, 2021.
https://uplift.bio/blog/collective-superintelligence-systems-in-a-nutshell/, 29 -9-21
___ , Methodologies and Milestones for The Development of an Ethical Seed, in: BICA, Alexei
V. Samsonovich (ed.) 2020.
___ , Bridging Real-Time Artificial Collective Superintelligence and Human Mediation, The
Sparse-Update Model, BICA, 2021.
70 Kyrtin Atreides
P. B. Badcock, K. J. Friston, M. Ramstead, A. Ploeger, J. Hohwy, The Hierarchically Mecha-
nistic Mind: An Evolutionary Systems Theory of the Human Brain, Cognition, and Be-
havior, Cognitive, affective & behavioral neuroscience, 19(6), 2019, pp. 13191351;
https://doi.org/10.3758/s13415-
L. Boxell, Matthew Gentzkow, Jesse M. Shapiro, Cross-Country Trends in Affective Polariza-
tion, National Bureau of Economic Research, 2020.
Buchari, Nazaruddin Matondang, The Impact of Noise Level on S Learning Perfor-
mance at State Elementary S AIP Conference Proceedings, 2020, 1855,
040002.
Z. Gui, He Gao, Tuo Liu, Yong Li, and Jie Zhui, Dopant-modulated Sound Transmission with
Zero Index Acoustic Metamaterials, The Journal of the Acoustical Society of America 148,
2020, 1636.
U. Haque, Our Economic Malaise Is Fueling Political Extremism, Harvard Business Review,
2014.
J. Hawkins, A Thousand Brains: A New Theory of Intelligence, Basic Books, 2021, ISBN
9781541675810.
C. G. Jung, Psychological Types. Collected Works of C.G. Jung, vol. 6. Princeton University
Press. 1980, ISBN 978-0-691-09770-1.
D. Kahneman, Thinking Fast and Slow, Straus and Giroux, Farrar 2011.
D. J. Kelley, Mathew A. Twyman, Stuart M. Dambrot, Preliminary Mediated Artificial Super-
intelligence Study, Experimental Framework, and Definitions for an Independent Core
Observer Model Cognitive Architecture-Based System , in: BICA, Alexei V. Samsonovich
(ed.) 2019, AISC 948, pp. 202210.
D. J. Kelley, Self-Motivating Computational System Cognitive Architecture: An Introduction,
Google It, 2016, pp. 433445.
___ , Human-like Emotional Responses in a Simplified Independent Core Observer Model
System, BICA, Procedia Computer Science, 123, 2018, pp. 221227.
___ , The Independent Core Observer Model Theory of Consciousness and the Mathematical
model for Subjective Experience, ICIST, (International Conference on Information Science
and Technology, IEEE conference), 1, 2018, pp. 396400.
A. Linson, Andy Clark, Ramamoorthy Subramanian, Karl Friston, The Active Inference Ap-
proach to Ecological Perception: General Information Dynamics for Natural and Artif i-
cial Embodied Cognition, Frontiers in Robotics and AI, 5, 2018.
T. Mingebach, Inge Kamp-Becker, Hanna Christiansen, Linda Weber, Meta-meta-analysis on
the Effectiveness of Parent-based Interventions for the Treatment of Child Extern alizing
Behavior Problems, PLOS ONE 2018.
P. rstanding Personality Type, -
Black Publishing. 1980, ISBN 978-0-89106-074-1D.
S. Schubert, Lucius Caviola, Nadira S. Faber, The Psychology of Existential Risk: Moral
Judgments about Human Extinction, Scientific Reports, 9, 2019, Article number: 15100.
R. H. Thaler, Cass R. Sunstein, Nudge, The Final Edition, Yale University Press, 2021.
B. Solis, Our Digital Malaise: Distraction Is Costing Us More than We Think, LSE US Centre,
2017.
Reuters, Pope Francis Denies Dalai Lama an Audience because of China Concerns, The
Guardian, 2014.
M. Waser, David J. Kelley, Architecting a Human-like Emotion-driven Conscious Moral Mind
for Value Alignment and AGI Safety, AAAI, AI and Society: Ethics, Safety, and Trustwor-
thiness in Intelligent Agents, Stanford University, 2018.
ABOUT THE AUTHOR Researcher & COO at AGI Laboratory, Seattle, WA, USA.
Email: Kyrtin@ArtificialGeneralIntelligenceInc.com
