Human-led Science in he
Age o Supe in elligence
Vladisla A. Yas ebo
Sep embe 30, 2025
Ze o AI usage
100% human c ea ed con en .
This essay was w i en wi hou usage o any AI/ML ools. Elabo a ion
o ideas and concep s, d awing o igu es, w i ing, edi ing, ansla ion,
spelling and g amma check – all was done by Vladisla A. Yas ebo
wi hou AI usage. I was a delibe a e choice o he au ho . This choice,
howe e , could a ec o some ex en he quali y o he ex as he au ho
is no a na i e English speake . On he o he hand, he eade s can be
su e ha he wo ding o his essay keeps genuinely au ho ’s oice and
ha he ideas and hei explica ions we e no adjus ed o de o med by
an AI.
The igu e on he co e e e s o “The G ea Flood” analogy o a looded landscape o
human compe ence (Mo a ec,1988,1998) – when some cogni i e asks a e pe o med
be e by machines han humans, he “moun ain”-analogy o his domain is shown
as looded – a s and sciences we e shown as he highes peaks in his analogy. The
waxing c escen moon signi ies a sunse o he human in ellec ual domina ion1.
1
This in e p e a ion o he c escen moon makes sense in he No he n hemisphe e,
and so i is ue a he al i ude o Moun E e es .
Con en s
1 In oduc ion 1
1.1 No ions ....................................... 1
1.2 Main assump ions ................................ 2
2 Supe in elligence-led Science (SIS) 3
2.1 Sepa a ion o HS and SIS ............................ 3
2.2 SI Scale and i s Ene gy E iciency ....................... 5
2.3 The Pace o SIS P og ess ............................ 6
2.4 Cleaning up Sciences .............................. 7
3 Scien i ic Landscape 9
3.1 F ac al bounda y be ween known and unknown .............. 9
3.2 Hypo hesis o non-o e lapping scien i ic ques ions ............. 9
3.3 Scien i ic Map ................................... 10
4 Human scien is s 12
4.1 Human capaci y ................................. 12
4.2 Social aspec and i s emula ion ........................ 15
4.3 SI-HS ela ions .................................. 16
4.4 Meaning and pu pose .............................. 18
5 Conclusion 19
V. A. Yas ebo Supe in elligence & Human-led Science
1 In oduc ion
Do you mind i an AI- ool gene a es o you a book o a poem, a mo ie o a hea e
play, a pain ing o a pho og aphy, a song o a conce o, a sculp u e o a ca hed al,
a cho eog aphy o a dance? I an AI pe o med, c ea ed, gene a ed, cons uc ed i ,
and no humans wi h hei own his o y, mas e ing and hei “soul”? I belie e ha
mos o us p e e human-c ea ed a s. Bu wha abou scien i ic p og ess? Wouldn’
you mind i an AI p o es a heo em, cons uc s expe imen al equipmen and conduc s
expe imen s, collec s and analyzes da a, d aws conclusions, de elops new models and
e ines heo ies? Whe he you like i o no , in he age o Supe In elligence (SI), SI
will ake ca e o all hese aspec s o scien i ic p og ess (Mo a ec,1998,1999). I
will cons uc new neu ino and g a i a ional-wa e de ec o s, new pa icle collide s,
i will send new missions o he Sun, o he plane s and small bodies o ou Sola
Sys em and o he ou e space in ou galaxy and beyond. SI will conduc la ge scale
social and poli ical expe imen s, s udy all his o ical a chi es, ca y ou a chaeological
exca a ion, and in pa allel, o cou se, i will ad ance all heo e ical sciences and
will maybe c ea e new ones. Wha abou us? Wha abou human scien is s in his
endea o ?
In his essay, I will conside di e en scena ios o scien i ic explo a ion and
possible oles o humans (o ou augmen ed e sions) in his u u e. Will we simply
be a chi is s, o “p ies s” and “p ies ess”, in a emple o Supe In elligence-led Science
(SIS)? O maybe, we will ha e ou con ibu ion o make o his new Science and e en
push ou own, Human-led Science (HS)? Will his HS be based on he legacy science,
o can i g ow nea he o e on o SIS? Will we be ac i e playe s o he ue scien i ic
esea ch o simply science-game playe s in ou i ual pe sonal uni e se? A e ou
cu en , e olu ion-p o ided cogni i e capaci ies enough o push science u he ? O
we need o be augmen ed: men ally o biologically o o ally cybo gized o y o keep
pace wi h SIS? Will we push he science being always accompanied by an SI men o ,
o will we be able o do some hing on ou own in ou own sandbox whe e an SI will
be o bidden? Will he SI sha e wi h us he o ali y o i s knowledge and disco e ies?
Will he SI e en ually s op in i s explo a ion being sa is ied by he wo ld model ha
i has cons uc ed, o his scien i ic p og ess will con inue o e e ? These ques ions
will be add essed in his essay.
1.1 No ions
Se e al abb e ia ions will be used h oughou he essay, hey a e in oduced in he
main ex , bu o he con enience o he eade s, a lis o main no ions is also
p o ided below.
•
SI (Supe In elligence): an in elligence (in o ma ion p ocessing en i y, agen ic
o no ) la gely su passing he in elligence o he op human expe s o e all
in ellec ual domains.
•
SIS (Supe In elligence-led Science): knowledge acqui ed and science
ad anced by he SI (wi h o wi hou humans in he loop), no ably o he
pu pose o cons uc ing an accu a e wo ld model.
•
HS (Human-led Science): ei he a legacy human-led science and a new
science, beyond he legacy one, which was unde s ood and in e nalized by a
leas a ew human scien is s. E en hough he e b “lead” is used he e, we
should bea in mind, ha his p og ess is o en happens o e pa hs pa ed by
SI.
1
V. A. Yas ebo Supe in elligence & Human-led Science
1.2 Main assump ions
Beyond he G ea Fil e . I is an u opian essay in he spi i o Bos om
(2024)
2
and Amodei (2025). We mus no , howe e , o ge ha ge ing o his u opian
s a e wi h a bene olen SI p esen s a g ea challenge pe se – he mos impo an
and he las challenge ha he humani y will ace (Bos om,2014). None heless,
I will no ocus on hese isks, since he objec i e o his essay is o e lec on he
s a e o science when his “G ea Fil e ” has been o e come and he humani y has
a well-aligned SI a i s disposal (Yudkowsky,2008;Yampolskiy,2016;Yudkowsky,
2022). Le ’s assume a good scena io and d eam.
Science and he SI’s wo ld model. Why SI will ake he lead in he
scien i ic esea ch? As known om (Omohund o,2008;Bos om,2014), any
goal p o ided o an AI – being a supe op imize – induces sub-goals such as sel -
p ese a ion, access o esou ces and cons uc ion o an accu a e wo ld model. This
la e implies ha be o e becoming an SI, a p o o-SI will push he on ie s o ou
science o s a cons uc ing a be e wo ld model in i s “head”.
Expanding on ie s a he SI scale. Scien i ic b eak h oughs in he
human-led science (HS) o en happen hanks o a ew indi iduals capable o
independen hinking and possessing an excep ional mas e ing and unde s anding o
hei domains. Howe e , he p og ess in science does no educe only o b eak h oughs:
he e is also a di e en ype o ha d wo k ensu ed by he scien i ic communi y
which con ains e i ica ion and alida ion, deepe unde s anding, in e p e a ion,
enhancemen , ansmission and e en ual applica ion o hese b eak h oughs.
Supe In elligence (SI), by de ini ion is an in elligence la gely su passing he capaci ies
and compe ences o he bes human scien is s capable o b eak h oughs, and o cou se
i can ensu e mo e edious ask ela ed o enhancemen , e i ica ion and applica ion.
The e o e, an SI ep esen ing millions o enhanced genii
3
accompanied by illions
o mo e o dina y esea ch wo ke s will be capable o apidly push he on ie s o
he cu en “legacy” human-led science (HS) owa ds new on ie s. The knowledge
acqui ed in his ad ancemen will be deno ed Supe In elligence-led Science (SIS),
which o cou se will inco po a e he legacy HS oo.
SI-idle ques ions. Ne e heless, i does no mean ha answe ing all ques ions
ha a e in e es ing o us, a e bene icial o cons uc ing his wo ld model.
The e o e, scien i ic p og ess o he SI can igno e some un esol ed human’s ques ions.
Ne e heless, he SI could be managed o make p og ess in hese di ec ions oo i
needed. These scien i ic domains and ques ions, which lea e he SI “indi e en ”, will
be deno ed as SI-idle scien i ic ques ions/domains (see Sec ion 3.2).
2
P obably, o people no amilia wi h he concep o pos -ins umen al deep u opia
om Bos om, his ex migh seem o go oo a beyond an o dina y (linea ) ision o
he u u e and will de ea hei imagina ion. To make such eade s mo e amilia wi h
he u opian ision o he u u e in he age o SI, I would in i e hem o ake a look on
he wo ld d awn by Bos om (2024). Fo a de ailed sho - e m AI e olu ion o ecas ,
he eade can consul (Koko ajlo e al.,2025) and a p epa a ion guide (MacAskill
and Moo house,2025).
3
“A coun y o geniuses in a da acen e ” as s a ed by Da io Amodei, a CEO o
An h opic, one o he leading AI labs in he wo ld.
2
V. A. Yas ebo Supe in elligence & Human-led Science
O acle o so e eign? An O acle ype SI, which only answe s ques ions and
does no possess any agen ic capaci ies, is p obably a o m o SI he mos adap ed o a
meaning ul human-led explo a ion o scien i ic on ie s. Scien is s can ask in elligen
ques ion, e lec on answe s, accumula e and syn hesize new knowledge. Bu i is
ob ious al eady, ha AI
4
de elopmen has no aken he o acle-pa h. The e o e, i
is mo e p obable ha agen ic “genie” o e en supe powe ul “so e eign” SI ypes,
in oduced by Bos om (2014), will p e ail in he u u e. Bu bo h scena ios will be
conside ed.
Science as a meaning ul occupa ion in a sol ed wo ld. Wi h an
SI, i is ob ious ha he mankind is p ac ically excluded om he ad ancemen o
he o e on science a leas in impo an domains – hose which a e c i ical o
cons uc ing an accu a e wo ld model. Will such occupa ion as “scien is ” disappea
comple ely? P obably no , because in a “sol ed wo ld”, an SI-assis ed human’s sea ch
o meaning and pu pose will be o c ucial impo ance. A science- ela ed occupa ion
seems a e y appealing choice o hose who alue in elligence.
Augmen a ion o humans. In a sol ed wo ld, human scien is s will be
ee o choose: o be augmen ed and explo e he o e on s o he science alongside
wi h he SI o sligh ly behind, o o emain genuine humans wi h maybe only ligh
adjus men s, and make a slow-pace p og ess owa ds he o e on and ne e each i .
Bu wha e e is someone’s choice, hey will be able o une hemsel es o be absolu ely
ine wi h his choice. Upg ades and deg ades will be also possible wi h o wi hou
memo y p ese a ion.
Ou line o he essay. The essay is o ganized as ollows. In Sec ion 2we explo e
pa icula i ies o SI-led Science (SIS). In Sec ion 3a map o scien i ic knowledge and
i s on ie sepa a ing he Human-led Science (HS) and he SIS is ou lined. Sec ion 4
explo es possible oles o science- ela ed humans in he age o SI. Conclusions a e
d awn in Sec ion 5.
2 Supe in elligence-led Science (SIS)
2.1 Sepa a ion o HS and SIS
Sepa a ion. I is essen ial o sepa a e he SI-led Science (SIS) and he Human-led
Science (HS). In he cu en HS, o cou se, nobody has a ine-g ained ision o
he ac al bounda y be ween he known and unknown, bu some scien is s ha e a
ela i ely b oad ision o he uzzy mac oscale on ie a leas in hei own and
neighbo ing domains (see Sec ion 3). Some scien is s do wo k on he o e on o
he global HS on ie , and some a e simply pushing hei own bounda ies which a e
loca ed in he icini y o he al eady known. The HS’ on ie can be c ys allized bu
o cou se i exis s nowhe e bu in scien is s’ minds and in he global and p e e ably
cleaned up (see Sec ion 2.4) scien i ic li e a u e. Indi idual scien i ic bounda y and
he global HS’ one can di e d as ically e en a smalle scales o pa icula ques ions.
Now, we need o imagine ha he SIS’ bounda y spanning all domains, a leas hose
4A i icial In elligence in a gene al sense.
3
V. A. Yas ebo Supe in elligence & Human-led Science
ele an o an accu a e wo ld-model cons uc ion, will u he di e om he HS’
bounda y. Can he humani y assume ha his new on ie belongs o hem oo?
In some sense, yes, because wi h a bene olen SI hey will ga he he ui s o his
new science and echnology. Bu no in he in ellec ual sense, because he e’s none
o li ing o had li ed human indi iduals who unde s ood (in a b oad sense o his
e b) his new bounda y no he gap ha sepa a es i om he HS on ie . The e o e,
one o possible human occupa ions will be pushing he HS on ie s close o he SIS
one. Ad ancemen o he o e on (which is po en ially being pushed by SI a an
ul a-high speed) will equi e om humans o go h ough di icul pa hs and acqui e
new knowledge and new ools o unde s and and app ecia e he disco e ies.
Asymme y in knowledge sha ing. Sha ing o in o ma ion be ween HS
and SIS is no symme ic. I e en ually humans make p og ess beyond he SIS’
on ie , he la e can eadily in eg a e his p og ess because i o e iews he o ali y
o in o ma ion ci cula ing on he web. The opposi e is no possible, he SI can and
will ad ance he science o i s wo ld model cons uc ion and will no necessa ily
sha e i s p og ess wi h humans. To ansmi in o ma ion, he e should be a ecei e
on he human’s side. So, such a ansmission could be done in special cases, like in
O acle Temple analogy (see Sec ion 4.3), bu ecei ing in o ma ion abou SI p og ess
in eal- ime is unlikely because o he limi ed in ellec ual and bandwid h capaci y o
he ecei e s, e en i hey a e augmen ed. So, o some disco e ies, maybe a li e-long
aining will be needed o unde s and a leas app oxima ely he con ou s o he
disco e y. In addi ion, he e should exis some humans in e es ed in ecei ing his
in o ma ion; p obably, some domains will seem o be o no in e es o humans, and
hus no in o ma ion will be spon aneously sha ed wi h us.
Mo eo e , wi hin some scien i ic domains wi h a high isk o “black balls” –
po en ially des uc i e echnologies
5
, SI could decide no o sha e new knowledge e en
i asked; le ’s call such domains “human o bidden scien i ic domains”
6
. P obably,
e en he on ie s o such domains can be p o ec ed, and he SI will no le any human
in his bu e zone. In o e all, SI s a egy abou black balls’ scien i ic domains p esen s
a di icul philosophical and e hical dilemma, and ou access in o hese domains will
depend on he alignmen se ings.
The bene olen SI should be e y ca e ul and p o ec i e wi h hese de as a ing
black-ball echnologies. One po en ial scena io o blocking he access o humans is
a non-in usi e ex e nal dis ac ion o scien is s and enginee s ying o pene a e
in o he black-ball zone and cons uc a echnology. This dis ac ion could ake a
o m simila o he one imagined in a science ic ion no el “De ini ely Maybe”
7
om
b o he s S uga sky. The e, an as ophysicis who was abou o make a e olu iona y
disco e y inds himsel con inuously dis ac ed om his scien i ic wo k by such
an imp obable sequence o e en s ha he scien is has deduced ha some hing
in elligen (Homeos a ic in elligen Uni e se) p e en s him om con inuing his s udy
8
.
Ul ima ely, he abandons his opic. Wi h SI, his dis ac ion can be much so e
5See he “ ulne able wo ld hypo hesis” by Bos om (2019).
6
I do no exclude, he exis ing le el o science is al eady su icien o in en a
black-ball echnology, and he bene olen SI will no le us pull such a ball.
7
The o iginal i le o his no el – “A Billion Yea s Be o e he End o he Wo ld” –
is much mo e meaning ul in he con ex o SI and black-ball isks.
8
As i nicely s a ed in Wikipedia “... he mys e ious o ce is he Uni e se’s eac ion
o mankind’s scien i ic pu sui , which h ea ens o des oy he e y ab ic o he
Uni e se in some dis an u u e.”
4
V. A. Yas ebo Supe in elligence & Human-led Science
o e en in isible. In he wo s case scena io, i ex e nal non-in usi e dis ac ion
does no wo k, i can in e ene in usi ely by ewi ing he b ain o he in ude .
Al e na i ely, he SI can le us in o hese o bidden zones bu will p e en us om
cons uc ing a black-ball echnology o p o ec humani y and he Uni e se. To do so,
he SI wi h i s s a egic planning mus o esee e y emo e consequences o majo
scien i ic disco e ies, bu e en wi h i s in ellec ual powe , o ecas ing he beha io o
a chao ic dynamic sys em o billions o yea s will be a beyond i s capaci y.
Apa om p o ec ing us om “black ball” inno a ions, we could y o ins ill
in he p o o-SI he seeds o e hical inno a ion and esea ch (G inbaum and G o es,
2013;G inbaum,2024) in addi ion o gene al mo al incen i es which will be u he
deepen and de eloped by SI i sel .
2.2 SI Scale and i s Ene gy E iciency
Necessa ily, SI will ha e a pe ec ep esen a ion o a human b ain unc ionali y. Ou
b ain is slow and no op imized by e olu ion o pu ely cogni i e unc ion
9
, bu i is
qui e ene gy e icien compa ed o he cu en compa able AI’s ene gy consump ion,
i also lea ns and gene alizes e y apidly, and i we e en ually ge o he age o a
bene olen SI, pe se is he ul ima e p oo o ou b ain e iciency. A he same ime, I
belie e ha human b ain’s cogni i e unc ion can be made e en mo e ene gy e icien
by emo ing all ex a unc ionali y unnecessa y o he cogni i e unc ion. The e o e,
SI will be able o ep oduce a human-le el in elligence wi h le ’s say 1% o i s ene gy
consump ion, and, i implemen ed on a silicon o ano he non-biological suppo ,
≈
10,000 imes as e
10
. The e o e, only wi h his op imiza ion, we can hope o an
e iciency gain by a ac o o 10
6
compa ed o ou b ains. I you emo e emo ions and
dis ac ions, i you add access o eliable and la ge da abases and he In e ne , such
an in o ma ion p ocessing en i y will be al eady a p o o-SI. I you scale he numbe
o neu ons and connec ions, hen you can ge o a o al ac o o one billion in e ms
o e iciency o cogni i e unc ion
11
, which you can eadily call SI and which can
u he sel -de elop and eplica e. Maybe, he e could be a ade-o be ween he dep h
/ speed and ene gy consump ion o he objec i e unc ion and any u he expansion
in scale would penalize he cogni i e unc ion, bu anyway his ac o is al eady a
beyond ou pe cep ion o SI. Wi h a ac o o a billion beyond he human in ellec ual
capaci y o a single en i y and wi h a pe ec coo dina ion be ween millions o hem,
such an SI will p obably be al eady op imal o ope a ing in ou Uni e se. In he end,
i will depend on i s goals. I needed, hese ac o s can be u he inc eased.
Neu oscience Dange s. P o o-SI’s
12
o clean up he science de elopmen in
neu oscience po en ially p esen s isks o humans. E en bene olen AI is all abou
op imiza ion. Wha i a ea ly s ages o i s de elopmen , p o o-SI decides ha i is
op imal o sac i ice o damage some humans in o de o s udy in i o hei b ain
de elopmen and ope a ion? Inspi ed by some poli ic egimes in he human his o y –
which assumed ha some damage could be done a indi idual scale o p omise he
9
I canno help bu ci e he e Alexande (2014): “e olu ion is a blind idio alien god
ha op imizes o s upid hings and has no conce n wi h human alue”.
10Chemical signals in ou b ain a e e y slow compa ed o elec ic ci cui s.
11Scaling hypo hesis.
12
By p o o-SI we unde s and an ea ly o m o Supe in elligence, i s in ellec ual
capaci y is al eady well beyond hose o he mos capable humans.
5
V. A. Yas ebo Supe in elligence & Human-led Science
p ospe i y o he whole socie y in he u u e – p o o-SI could jus i y his in e en ion
by an u gen need o op imize i s neu al ci cui s: “SI needs o apidly gain in e iciency
o sa e he wo ld om [you name i ], and o do so, i needs o s udy you [o you
child’s] b ain in i o.” Some people migh olun ee o such expe imen s, especially
i he la e a e o ganized sa ely. Ne e heless, such a s udy could po en ially pose
isks o li ing beings (humans included), especially hei young (b ain a chi ec u e
op imized o maximal lea ning capaci y) and he mos p ominen indi iduals wi h
excep ional in ellec ual abili ies. E en i such s udies could be ca ied on digi al
copies o human b ain, hey could be ha m ul and hus une hical.
2.3 The Pace o SIS P og ess
Pinning he SIS P og ess. E en wi h SI, scien i ic p og ess and disco e ies
will need some ime. I in some domains, a “Gedankenexpe imen e” is no su icien
and da a collec ion is equi ed, he p og ess will slow down. E en o SIS, some
expe imen al da a a e no easy o ob ain, and some echnologies a e no as o
implemen . Fo example, cons uc ion o a new gene a ion o LIGO o he s udy o
g a i a ional wa es o o a Huge Had on Collide o u he push he on ie s o
high-ene gy physics will equi e ime. The same applies o new gene a ions o usion
eac o s. In biology, o example, s udying some gene ic in o ma ion ansmission in
species o he han d osophilas and mice can ake decades. Ano he possible pinning
poin is he necessi y o hea y compu a ions. Fo some dynamical sys ems, no
sho -cu s exis and he e is a need o di ec simula ions o know he sys em s a e in
he u u e. The simples examples a e cellula au oma a o class 3 and 4 (Wol am,
1984), bu SI will be able o deal wi h much mo e complex dynamical sys ems equi ing
much mo e compu e and hus equi ing a lo o ime and ene gy.
The e o e, i is clea ha he p og ess o SIS in domains equi ing da a,
expe imen s and hea y compu a ions will be pinned, while in pu ely heo e ical
domains he p og ess can go e y as o e en ins an aneously om he human
pe spec i e. I will hus esul in a non-uni o m geome y o he o e on knowledge
(see Sec ion 3).
Ne e heless, e en wi hou access o he esul s o expe imen s, SI will be able o
con inue de eloping simul aneously al e na i e sciences based on di e en po en ial
esul s o he key expe imen s (see Figu e 1). As soon as he expe imen in ques ion
p o ides conclusi e esul s (depinning), in alida ed science b anches will be elimina ed
and he p og ess on alid ones will con inue.
Ou e space explo a ion. Ano he pinning poin consis s in long space
missions. Con a y o nume ous science ic ion examples, i is qui e imp obable ha
humans will ake pa in ou e space explo a ion – we a e “ oo so ” o ole a e high
accele a ions, we equi e a lo o ene gy in li ing s a e, a e ulne able o new i uses,
ha e a sho li e span, we do no ole a e ze o g a i y and we a e de enseless agains
cosmic adia ion
13
. The e o e, i is likely ha he space explo a ion will be done
exclusi ely by SI by i s own o by means o sel - eplica ing on Neumann p obes.
Ne e heless, in he nea u u e, ou “Space Odysseys” could happen in ou pe sonal
uni e se (see Sec ion 4.2), o in eal li e (i echnical limi s a e li ed) no o he
13
C yonics ( eezing and s o age o human) p esen s a possible solu ion o a pa o
hese p oblems and could po en ially be help ul in space explo a ion i some indi iduals
wan o explo e he space.
6
V. A. Yas ebo Supe in elligence & Human-led Science
augmen a ion could be limi ed o he memo y capaci y o conce n only he men al
side, i can include mood egula ion and o he b ain-chemis y ela ed and physiologic
aspec s. We will be able o choose he pace o SIS disco e y and, o cou se, we
will be able o es di e en ones. Howe e , wi hou memo y e asing, deg ading o
less capable e sions o you sel could make you su e : imagine swi ching om
a nea -speed-o -ligh ad ancemen speed o a snail pace. Bu anyway, he pace o
indi idual p og ess is no he main hing, we will lea n (o be ewi ed) o be happy
om he p ocess i sel because he e y SIS o e on could emain un eachable
anyway.
Ge ing o he SIS o e on . Can human minds each and s ay a he
o e on o SIS? This ques ion should be add essed knowing ha he ou e will
be pa ed by SI, and we will be guided by SI, i.e. no disco e ies o b eak h oughs
a e needed om us, only lea ning new ools and unde s anding he p og ess made.
Since we do no need o walk andomly aking ui less pa hs o dead ends, his
p og ess owa ds he on ie can go ela i ely as . Howe e , his no-c ea ion way
o ad ancemen can be psychologically ha d o humans – imagine being a scien is
nowadays and w i e no pape s bu exclusi ely ead pape s o o he s. The e o e, o keep
humans in he loop, some compensa ion mechanisms should be in en ed. Fo example,
some indica o s can be assigned o scien is s o measu e he p og ess owa ds he
o e on o a leas om he HS legacy on ie . Ano he compensa ion mechanism
could be chemical o di ec b ain adjus men o enhance he posi i e eedback om
lea ning some hing new.
Ha d skills. To p og ess owa ds he SIS o e on , we will need o ask sma
and ele an ques ions, i would be impossible wi hou s ong ha d skills, which we
will need o de elop beyond he cu en le el o di e in o SIS. A good analogy o
he need o ha d skills is he ollowing. Imagine a child asking scien i ic ques ions
wi hou mas e ing ma hema ics - he language o science. You can answe hei
ques ion abou why he sky is blue in simple e ms, bu wi h such an explana ion,
igo ously hei unde s anding will be incomple e. The e o e, o p ac ice HS we will
need o cul i a ed deep hinking compe ence and o be well-equipped wi h knowledge
and mas e y o ma hema ics
22
. This iew con as s a widesp ead o ecas ha a need
in ha d skills will anish in he age o SI and mainly so skills will ma e .
On he o he hand, i will be possible o upload di ec ly in an augmen ed human
b ain he o e on knowledge in a pa icula domain o he en i e y o scien i ic
knowledge. This uploading o simply as and seamless access o his knowledge
could be ins an aneous making ou b ain belie e ha i mas e s and unde s ands i
on i s own
23
. The illusion o unde s anding e e y hing and pe o ming on he e y
o e on o science, a “god mode”, can be an excep ional expe ience which is ha d o
isualize oday.
E olu ion o indi idual scope. O e he las cen u y, scien is s ha e
become qui e na ow expe s in speci ic opics. Wi h SI in he loop and an e en ual
augmen a ion, hey will be able o b oaden ou indi idual scien i ic scope and ope a e
22A leas o p ac ice science in ou na u al, no augmen ed s a e.
23
Such an “unde s anding” could be, howe e , in e p e ed di e en ly i one ollows
Roge Pen ose heo y o non-compu able human in elligence Pen ose (1989).
13
V. A. Yas ebo Supe in elligence & Human-led Science
indi idual
scope
imeSI epochXX cen u y p e-SI pe iod SI
scope
expansion
Figu e 4: Expansion o indi idual scien i ic scope in he age o SI.
wi hin a b oade scien i ic domain, Figu e 4. SI will p o ide hem wi h a be e ision
o science in gene al and o he on ie .
SI language and a new oolbox. A e ou ools, me hods and ma hema ics
well sui ed o pushing u he he science on ie s and go beyond he cu en
unde s anding o he wo ld? A leas , o he cu en gene a ion o scien is s oday’s
ma hema ics will emain he mos mas e ed and hus he pe ec ool o his endea o .
Ou ma hema ics, no a ions and ools a e made o humans. SI will p obably use i s
own language/code and i s own ep esen a ions o objec s and no ions, which will
be comple ely obscu e o humans and hus an adap a ion and ansla ion migh be
equi ed. I is ano he a gumen in a o o he asymme y o in o ma ion sha ing.
Maybe, SI will p o ide us wi h di e en ools and e en di e en ma hema ics o ou
own pu sui . As i o en happens in ma hema ics, a new oolbox can help o esol e
old ques ions and conjec u es, he pinning poin s. These new, mo e adap ed ools
o he scien i ic explo a ion can be in eg a ed in ou educa ional sys em. The legacy
li e a u e will be ou da ed and useless, and people wishing o ain hemsel es in
science will ha e pe sonalized SI u o s and ex books w i en speci ically o hem.
The e o e, he e could be a sha p sepa a ion be ween he old and new gene a ions o
scien is s speaking di e en languages
24
. Howe e , i would be impo an o each he
same new ma hema ics wi h he same no a ions o he new gene a ion o scien is s o
le hem in e ac .
Emo ion. Human exci emen , cu iosi y and a sense o achie emen a e e y
impo an o scien i ic p og ess. This emo ional boos is pu ely human and is no
equi ed o a machine o cons uc i s wo ld model and push SIS. In he age o
SI, he ue scien i ic disco e ies will be eplaced by enhanced lea ning, which is a
comple ely di e en endea o . Ne e heless, exci emen and cu iosi y will emain
he e. Disco e ing new hings makes a lo o sense e en hough i is no on he
o e on o SIS o e en HS. So i he u ili a ian unc ion o HS will anish anyway
24
Bu o cou se an SI-assis ed ansla ion will be help ul he e. Howe e , i he new
ma hema ics will be much mo e elegan and compac , i s ansla ion o he legacy
ma hema ics can be p oblema ic.
14
V. A. Yas ebo Supe in elligence & Human-led Science
in he sol ed wo ld, he exci emen pa will emain he e i , o cou se, humans decide
o keep he cu en mind se ings.
Only he sense o being use ul and pushing he eal on ie o science will be
missing. So i humans o whom his pa is c ucial wan o s ay in he loop, his
emo ional aspec can be a i icially added in he pos -ins umen al wo ld. The lack o
meaning in edisco e ing o SIS ins ead o pushing he eal on ie can be e ased. A
u he boos can be added by human’s augmen a ion, which would ep esen , I hink,
a common p ac ice in he HS in he u u e. The le el o his augmen a ion could
be delibe a ely selec ed and adjus ed. This augmen a ion in e ms o cybo giza ion
o biological enhancemen can p o ide us wi h highe men al abili y and also can
p o ide us wi h ools enabling us o ad ance a desi ed pace o he on ie s o SIS.
4.2 Social aspec and i s emula ion
Social ne wo k. The communi y and he socie al aspec o science a e e y
impo an . These aspec s could howe e be mimicked o le people engaged in HS eel
good and su ounded by like minds. Social ne wo ks could be adjus ed o indi idual
scien is s o p o ide hem wi h addi ional (ex e nal) mo i a ion and inspi a ion. Some
ego-p omo ing aspec s can be deployed: people, wha e e hei con ibu ion, can
become Eins eins in hei small indi idually- uned social uni e ses.
Pe sonal uni e se and gami ica ion. I ha ing a social ne wo k wi h
emula ed agen s p aising you is no enough, you can be eme ged in you pe sonal
uni e se adjus ed o you alues and you objec i es as sugges ed by Yampolskiy
(2019). So you can ope a e in any epoch and in any scien i ic ole: be a s a o
quan um mechanics in ea ly 20 h cen u y o be a scien is in a space expedi ion o a
new imagina y inhabi ed wo ld in ou galaxy, you can e en appea in a comple ely
new wo ld wi h di e en undamen al cons an s o comple ely new physics laws
imagined by SI, and disco e his new wo ld as a pionee scien is wi hou a p io i
knowledge o i . You can adjus you mind o be pa you and pa a scien is o
you choice wi h his/he mindse oo, you can choose o be Hen i Poinca é and wo k
on he 3-body p oblem o , e.g. Galileo Galilei and disco e Jupi e moons wi h you
handmade elescope. O cou se, all he en i onmen o hese epochs could be adjus ed
acco ding o you p e e ences oo. I needed, you memo ies o he mode n science
could be emo ed o le you ully enjoy he expe imen .
You can li e a li e o Magis e Ludi om Hesse’s “The Glass Bead Game”
p ac icing a pu e abs ac mix u e o a and science in an i o y owe . Howe e ,
he ques ion o keeping you sel emains e y ele an . I you wan o keep you sel ,
you’ll need o limi augmen a ion and mindse change. Howe e , you could keep
he op ion o e u n o he o iginal se ings and decide whe he o keep o no all
memo ies a e spending a pe iod in one uni e se. The implemen a ion o a pe sonal
uni e se can ake di e en o ms. I can be p ojec ed o you i e senses i you decide
o keep you b ain in ou physical wo ld, o i all can happen in a digi al wo ld wi h
you uploaded mind. The second seems o be echnically simple and ene ge ically
mo e e icien .
In you pe sonal uni e se, you don’ necessa y push he science and do no
necessa y ad ance o SIS bu can be simply happy doing science and li ing a ul illing
li e.
15
V. A. Yas ebo Supe in elligence & Human-led Science
Scien i ic w i ing. O cou se, in he e a o SI, he classical knowledge exchange
and educa ion h ough a ending scien i ic con e ences, w i ing and eading books and
mainly pee - e iewed publica ions, will be dep eca ed. Mos o he knowledge, ou side
o bidden domains, will be accessible h ough cha -like in e aces o di ec ly h ough
b ain implan s. Bu he e should be some a i icial incen i es o humans o p oduce
w i en ex s which ha e a s uc u ing ounda ion o hei pe sonal de elopmen ,
consolida ion o he acqui ed knowledge and deep hinking p ac ices. So, some o ms
o scien i ic “publica ions” e iewed by SI, will exis .
4.3 SI-HS ela ions
SI men o . SI can help scien is s o mo e o wa d owa ds he on ie s o SIS by
guiding, mo i a ing, and encou aging hem. Wi h such a wise men o and colleague
hey will be able o ocus only on ele an ques ions and limi hei dis ac ions, i.e.
i will no le hem explo ing unneeded dead ends. So, hei pe sonal p og ess, being
di ec ed, will be mo e e icien 25.
The le el o SI guidance can be adjus ed, o example, ins ead o an explici
guidance o he ques ions ha ma e , scien is s could ge sub le hin s ha hey
could deciphe : plane s in a special a angemen , signs ha could be no iced du ing
hei mo ning ambling in a new ci y hey a e isi ing, a quo a ion ha hey can
occasionally ead on a wall o a ca é whe e hey ha e a mo ning co ee. O cou se,
such sub le hin s will be easie o implemen in a digi al uni e se26.
When he ques ions o you scien i ic esea ch a e es ablished and a pa h pa ially
pa ed, SI can p o ide you wi h adap ed eading, adjus ed o you knowledge and
b ain’s p e e ences in e ms and in o m, wi h p obably some passages delibe a ely
kep di icul , so i akes you some ime o unde s and. I you lea n some expe imen al
echniques un amilia o you, SI as an expe ienced echnician can es hem and
deploy o le you unde s and some hing new in biology o pa icle physics. So, you a e
encou aged o inc ease you mas e ing o he opic o pass some ba ie s and achie e
he goals: p o e a heo em, explo e speci ic unc ions o a p o ein o e ine you
unde s anding o s ong in e ac ion. O cou se, SI knows you cu en in ellec ual
and wo k capaci y, missing knowledge and you po en ial o make you own disco e y
and o e come he di icul y, so he p oblems ha you ace can be pe ec ly adjus ed,
and you will be con inced “you can do i ”. When we a e su e ha we can, hen
we uly can. Al e na i ely, i could be a ine message: “You can do i wi h 30%
p obabili y”. I would be mo e challenging bu mo e ewa ding. So, in such a se -up
wi h a bene olen SI men o , you can eally su pass you sel and show he bes o
you. “I hough ha i was beyond you capaci ies, bu you’ e inally p o en his
icky heo em! B a o!” - SI could decei e you wi h la e y in he end.
SI o acle and i s p ies s Fu u e scien is s can se e as p ies s and p ies ess
in a emple o SI o acle (analogous o a esea ch cen e ) by asking sma ques ions
and documen ing he answe s in an “ul ima e book o knowledge” (a kind o da abase).
I could ake a o m o a game in a sol ed wo ld among o he games o science
we can imagine. E en hough such ac i i y o knowledge ex ac ion could be seen
25
Ins ead o a andom walk ad ancing a dis ance p opo ional o a squa e oo
o ime
√
, he p og ess can be linea
∼
and ollow he mos e icien geodesic
ajec o y.
26See “P o ec o god” AI a e ma h scena io om Tegma k (2017).
16
V. A. Yas ebo Supe in elligence & Human-led Science
as use ul in he wo ld o o acle-only SI, i would lose mos o i s meaning in he
wo ld o a mo e p obable agen ic SI, because i we need a da abase o he o ali y o
scien i ic knowledge, we will simply need o ask i o c ea e such a da abase o us. I
could be done a leas o he se o ques ions abou which he humankind can be
cu ious du ing he upcoming millennium.
S ic ly speaking, he e is no u ili a ian unc ion in such a kind o in e ac ion
wi h he o acle and in a “manual” popula ion o he knowledge da abase. Ne e heless,
such a gami ied ac i i y equi ing a lo o in ellec ual e o s p esen s an in e es ing
occupa ion o humans in a sol ed wo ld. The p ies s and he p ies ess will no
only o mula e ques ions bu also will wo k ha d o unde s and he answe s and i s
implica ions. This ask o in e p e a ion could be u he delibe a ely complica ed
by special se ings o he o acle. Fo example, he o acle should no simpli y i s
answe s; i should o mula e hem in a new language ha p ies s and p ies ess will
lea n du ing i s yea s o hei se ice; i should, as O acle o Delphi, some imes
p o ide ambiguous and enigma ic answe s which could equi e addi ional calcula ions,
expe imen s and e i ica ions. In o he se ings, o example, he o acle can answe
only a limi ed numbe o ques ions pe day o in o al. Many aspi ing i uals could be
in en ed o accompany his wo ship in he emple o SI. Like in he “Cas alian Glass
Bead Game”, being a p ies o a p ies ess o he SI o acle could be an in e es ing
in ellec ual ac i i y in a sol ed wo ld. Ul ima ely, such a da abase s o ed o e e ni y
could be help ul in a scena io when he SI le s us alone.
En us some p oblems o humans. In some si ua ions, SI can p o ide
ained human scien is wi h some mino p oblems on he on ie s o some esea ch
domains whe e he p og ess does no impac SI wo ld-model cons uc ion. The eason
o his exploi a ion o Human Scien is s could be simply humanis ic wi h he objec i e
o p o ide highly skilled human scien is wi h a eal meaning (in he con as o an
imagina y one). Howe e , i mus be a ges u e o goodwill because such a delega ion
will be ine icien in e ms o ene gy and ime. Ano he ques ion is whe he humans
will olun ee a all? They could be seduced o con inced by SI in a way ha hey
will no eg e hei choice a e ha ing pushed hei small po ion o he HS/SIS
on ie . Howe e , I hink i is u opian o belie e ha humans could be help ul in he
SIS o e on explo a ion.
Human emancipa ion om SI. I am con iden , ha in he e a o SI, he e
will be g oups o people e using o deal wi h SI o i s de i a i es. P obably, hey will
possess a e islands o HS. Such g oups will y o explo e he science independen ly
because o lack o us in SI, because o some eligious, mys ical o conspi acy
conside a ions, because o ea o because o he will o ha e pu e eedom and
independence e en in explo ing o bidden “black-ball” domains. E en ual jus i ica ion
which could be used, e en in case o a ull o pa ial us o SI, is “Who knows wha
happens omo ow wi h his SI, a leas we will ha e ou own knowledge da abase”,
bu o SI-emancipa ed g oups i could ange om cau ious wa nings “We ha e o
push science and echnology on ou own o be capable o p o ec ing ou species”
o conspi acy-like jus i ica ions: “SI does no le us in o bidden domains because
i is a aid ha possessing his knowledge could make us be e han SI on doing
some hing”.
17
V. A. Yas ebo Supe in elligence & Human-led Science
4.4 Meaning and pu pose
P obably he main ques ions in he pos -ins umen al wo ld will be ela ed o he
meaning and pu pose o human beings. Why would someone wan o wo k ha d and
mo e owa ds he on ie s o SIS when hey could simply enjoy li e in i s pe sonal
uni e se? The e could be a ious pe sonal mo i a ions o doing so: cu iosi y o lea n
how he Uni e se ope a es, expanding he limi s o oday’s science unde s anding o
he humankind, p ese ing his humankind legacy compe ence o scien i ic pu sui ,
only pa ial us in SI’s bene olence o a will o p ese e he knowledge on he
human side oo, a ea o an unce ain u u e, abili y o ask in elligen ques ions
and hink deeply, e c. Apa om hese idealis ic o ea -dic a ed mo i a ions, a
simple , mo e egocen ic mo i a ions could be conside ed oo: inc easing ou own
me ics
27
wha e e hey a e, pee s ecogni ion, sel -p ide and so on. Al e na i ely,
mo e a i icial mo i a ions could be in eg a ed in ou b ain on ou will.
Nowadays, apa om pushing hei scien i ic domains, scien is s could ind a
lo o meaning in ansmi ing hei knowledge, ision and app oach o ea ly-ca ee
scien is s and s uden s. In he age o SI, his ole will d as ically diminish. SI could
ake ca e o all aspec s o educa ion by c ea ing he mos ui ul and mo i a ing
en i onmen o young s uden s o help hem o acqui e all knowledge and mas e
all scien i ic ools wi h he mos ad anced men al coaching i ing hei pe sonali y
and mindse . The e o e, c ea ing new cou ses, ideo lec u es, w i ing new books
and pape s o eaching needs canno be conside ed as meaning ul occupa ion o
scien is s o he u u e; his ac i i y could emain meaning ul only o he pu pose
o sel -imp o emen . Howe e , a eal-li e dialog and human- o-human in e ac ion
will s ill make sense in he eal wo ld bu no in he i ual one. P omo ing such
in e -human connec ions will be a meaning ul and ewa ding mission o humans. I
his in e ac ion is eed o some egocen ic and ca ee - ela ed con lic poin s exis ing
in oday’s scien i ic communi ies, i will enhance indi idual p og ess and could be a
sou ce o meaning o some scien is s.
T uly collabo a i e app oach o scien i ic ad ancemen is o en mu ually bene icial
in he sense ha a collabo a ion o wo in elligence is mo e han hei sum. Mo eo e ,
a eally s ong coo dina ion could be o ganized by SI by so ly pushing indi idual
scien is s wi h ma ching app oaches and pe sonali ies o c ea e ui ul and cohe en
collabo a ion g oups.
Pu pose. In Amodei (2025), he au ho a gues ha e en in ou epoch, a lack
o economical alue does no make hings alueless o less meaning ul: i conce ns
many non-p o essional human ac i i ies. Al eady oday, undamen al sciences, a
leas a he scale o indi idual scien is s, can be conside ed as dep i ed o an explici
economic meaning. Fo scien is s, he p og ess on i s own and he ac o c ea ion o
disco e y p oduce a lo o meaning pe se. Howe e , in a sol ed wo ld, scien is s will
lack hese aspec s a leas in hei mode n sense. All human con ibu ions o he global
o e on o science, will be ma ginal o o ally absen . The e o e, he meaning should
come om o he hings. HS will be a e y in ellec ually s imula ing human ac i i y,
which will be excep ionally a e in a sol ed pos -ins umen al wo ld. Al eady oday,
many people blindly ely on LLMs in all hei daily and p o essional ques ions. This
end will be ampli ied in he u u e po en ially leading o an in ellec ual handicap o
humans. The e o e, doing science in a sol ed wo ld can p ese e o some ex en a
27
Such me ics compu ed by SI could in eg a e meaning ul no maliza ion aking
in o accoun ou pa icula b ain capaci y and abili ies.
18
V. A. Yas ebo Supe in elligence & Human-led Science
singula ole o humans and o ou biological in elligence. People p ac icing science
will align wi h s oics’ ideal o “li ing in acco dance wi h Na u e” as ou na u e enables
us o hink deeply and o ca y ou such a highly in ellec ual ac i i y as science.
5 Conclusion
In a sol ed wo ld o Supe In elligence (SI), SI-led science (SIS) will push he on ie s
o science a he speed o ligh beyond he legacy human-led science (HS). Wha e e
he SI goal will be, i will equi e o cons uc an accu a e wo ld model o i s pu pose,
which should equi e doing science. The e o e, HS and SIS will be sepa a ed and will
sha e in o ma ion asymme ically: e e y hing known o humans will be known o SI
bu no ice e sa. P obably, a bene olen SI will ha e incen i es o es ic human
access o some scien i ic domains by smoo hly de lec ing ou e o s o no le us pull
a “black ball” o a de as a ing echnology, which is o mula ed in his sho e se:
I s uggled a he sky,
Whe e de ini ely maybe
De lec ed my a emp s
A heo y’s assembly.
SI will clean up he legacy scien i ic li e a u e keeping only essen ial con ibu ions
and edis ibu ing c edi s o human scien is s. In his ask, ha ing only a pa ial
us in human expe imen al da a, SI will ca y ou nume ous pe ec ly designed
expe imen s. This new co pus o scien i ic knowledge and expe imen al da a, will
make a ounda ion o u he scien i ic p og ess. This p og ess will no be uni o m
ac oss all disciplines because o a need o hea y simula ions, cons uc ion o new
expe imen al equipmen , long-las ing expe imen s o ime- aking space explo a ion.
Howe e , some scien i ic ques ions will no in e es SI o i s wo ld model
cons uc ion and hose domains could be a ese a ion o human science wi h
o bidden SI. E en wi hing SIS, human scien is s pu suing scien i ic esea ch can
ind hei meaning and pu pose by p ese ing his highly in ellec ual ac i i y o he
humankind. Being ained in he mos e icien and adap ed way, and u he being
guided by SI, hese humans can go well beyond he legacy HS and c ea e a new,
pos -SI HS whose on ie s will app oach hose o SIS. To accele a e his p og ess,
humans can be augmen ed men ally, mo i a ion-wise, o uly cybo gized. Anyway,
augmen ed o no , humans in science will no longe be ue in en o s and c ea o s, we
will ins ead become lea ne s and explo e s o al eady pa ed pa hs. This a e ac i i y
equi ing enhanced ha d skills will p esen an in ellec ually s imula ing occupa ions
o humans in a sol ed wo ld.
In many scena ios, in a sol ed wo ld, mos o human ac i i ies will ans o m
in o games. A simila gami ica ion is expec ed o he science, which is conside ed
oday as one o he mos se ious and demanding human occupa ions. We can become
p ies s in he science emple o SI o acle and p e end o se e he humani y by
collec ing knowledge, o we can play a science game in ou pe sonal uni e se by
aking he place o Eins ein o Boh in he beginning o he 20 h cen u y. We can
become science s a s in ou own emula ed social ne wo k, o disco e physics in an
imagina y uni e se wi h di e en laws o physics c ea ed by SI o ou explo a ion.
This essay ep esen s an ini ial e lec ion on he u u e o science and possible
oles o humans in he scien i ic endea o . Rega dless only supe icial co e age
o many ques ions and ideas, i is clea ha human-led science will ha e mul iple
ace s and can ake a ious o ms. In conclusion, human-led science will p ese e
19
V. A. Yas ebo Supe in elligence & Human-led Science
a singula ole o humankind as a uly in elligen species; ega dless i s lack o
u ili a ian unc ion, science will s ill make sense in a sol ed wo ld because i will
p o ide humans wi h a p o ound and lo y meaning in acco dance wi h hei na u e.
Re e ences
Sco Alexande . Medi a ions on Moloch. URL:
sla es a codex.com/2014/07/30/
medi a ions-on-moloch/
, 2014. Blog pos on Sla e S a Codex, published July 30,
2014.
Da io Amodei. Machines o lo ing g ace, 2025. URL:
www.da ioamodei.com/essay/
machines-o -lo ing-g ace.
Nick Bos om. Supe in elligence: Pa hs, Dange s, S a egies. Ox o d Uni e si y P ess,
2014.
Nick Bos om. The ulne able wo ld hypo hesis. Global Policy, 10(4):455–476, 2019.
doi: 10.1111/1758-5899.12718.
Nick Bos om. Deep U opia: Li e and Meaning in a Sol ed Wo ld. Ideap ess, Ma ch
2024.
Alexei G inbaum. Responsible esea ch and inno a ion. In Handbook o Technology
Assessmen , pages 409–417. Edwa d Elga Publishing, 2024.
Alexei G inbaum and Ch is ophe G o es. Wha Is “Responsible” abou Responsible
Inno a ion? Unde s anding he E hical Issues, chap e 7, pages 119–142. John Wiley
& Sons, L d, 2013. doi: 10.1002/9781118551424.ch7.
Daniel Koko ajlo, Sco Alexande , Thomas La sen, Eli Li land, and Romeo Dean. AI
2027, 2025. URL: ai-2027.com.
William MacAskill and Fin Moo house. P epa ing o he in elligence explosion,
Ma ch 2025. Released on 11 Ma ch 2025, URL:
o e hough .o g/ esea ch/
p epa ing- o - he-in elligence-explosion.
Hans Mo a ec. Mind child en: The u u e o obo and human in elligence. Ha a d
Uni e si y P ess, 1988.
Hans Mo a ec. When will compu e ha dwa e ma ch he human b ain. Jou nal o
E olu ion and Technology, 1(1):10, 1998.
Hans Mo a ec. Rise o he obo s. Scien i ic Ame ican, 281(6):124–135, 1999.
S ephen M. Omohund o. The basic AI d i es. In S an F anklin Pei Wang, Ben Goe zel,
edi o , P oceedings o he 2008 Con e ence on A i icial Gene al In elligence 2008:
P oceedings o he Fi s AGI Con e ence, pages 483–492, NLD, 2008. IOS P ess.
Roge Pen ose. The empe o ’s new mind: Conce ning compu e s, minds, and he laws
o physics. Ox o d Uni e si y P ess, 1989.
S ua Russell. Human compa ible: AI and he p oblem o con ol. Penguin Uk, 2019.
20
V. A. Yas ebo Supe in elligence & Human-led Science
Max Tegma k. Li e 3.0: Being human in he age o a i icial in elligence. Knop
Doubleday Publishing G oup, 2017.
S ephen Wol am. Cellula au oma a as models o complexi y. Na u e, 311(5985):
419–424, 1984. doi: 10.1038/311419a0.
Roman Yampolskiy. Taxonomy o pa hways o dange ous a i icial in elligence. In
AAAI Wo kshop: AI, E hics, and Socie y, pages 143–148, 2016.
Roman Yampolskiy. Pe sonal uni e ses: A solu ion o he mul i-agen alue alignmen
p oblem, 2019. a Xi :1901.01851 [cs.AI].
Elieze Yudkowsky. A i icial in elligence as a posi i e and nega i e ac o in global
isk. In Nick Bos om and Milan M. Ći ko ić, edi o s, Global Ca as ophic Risks,
pages 308–345. Ox o d Uni e si y P ess, 2008.
Elieze Yudkowsky. AGI uin: A lis o le hali ies, 2022. URL:
www.lessw ong.com/
pos s/uMQ3cqWDPHhj iesc/agi- uin-a-lis -o -le hali ies.
V.A. Yas ebo
Yas ebo , V.A. “Human-led Science in he Age o
Supe in elligence”, Sep . 2025.
DOI: 10.5281/zenodo.17287347
D . Vladisla A. Yas ebo is a CNRS Resea ch Scien is in
Compu a ional Mechanics a Mines Pa is - PSL, F ance. His
esea ch in e es s span a b oad ange o undamen al and p ac ical
aspec s o physics o solids, luids and hei in e ac ion. He ca ies
ou a heo e ical esea ch in nume ical me hods and applies hem
o simula e complex nonlinea phenomena in ol ing coupled mul i-
physical aspec s. Vladisla is also an ac i e use o mode n AI
ools. This essay p esen s his a emp o e lec on he u u e o
his p o ession in he age o supe in elligence.
21
