Roko

Roko Mijic was the kind of child who never stopped asking questions: how high is the sky? Does space have an edge, and if so what is outside it? Asking and answering the fundamental questions of our world is so much fun that I seem to be making a life of it.

The most interesting questions I have immersed myself in include category theory, the foundations of morality, evolutionary psychology, and my most recent passion: the likely trajectory of the future of intelligence and probably the entire universe - the technological singularity.

Education and work experience
2009 – present Volunteer, Singularity Institute for Artificial Intelligence
2008 – 2009 AI Researcher, University of Edinburgh, HP Labs Innovation Research Program
2007 – 2008 Certificate of Advanced Study in Mathematics, at St. John’s College, University of Cambridge
2002 – 2006 BA(Hons) in Mathematics and Physics at St. John’s College, University of Cambridge
Summer 2006 Summer Undergraduate Research Fellowship scheme at California Institute of Technology
2002 – 2005 (Holidays only) Oxford Instruments Superconductivity sponsorship scheme – trainee cryogenic engineer
1997 – 2002 Malvern College

Life-story

I was born in the UK in 1984 and spent my childhood near Dubrovnik in the southernmost tip of Croatia, Yugoslavia. My family was forced to move to England to avoid the onset of war in 1991, narrowly avoiding the siege of Dubrovnik. Upon arriving in England, I was lucky enough to obtain a government assisted place to a prestigious UK private school, Malvern College, where I was strongly influenced by some excellent teachers and by reading Wheeler and Taylor's Spacetime Physics and Hay's Quantum Universe - I decided that I wanted to be a famous physicist like Albert Einstein. In 2002 I was admitted to St. John's College, University of Cambridge to read mathematics and physics. After graduating with two firsts, two college prizes and the prestigious Certificate of Advanced study in mathematics, I decided to pursue my growing interest in the study of human intelligence and the future of humanity.

In 2008 I developed a serious interest in meta-ethics and transhumanism, influenced heavily by the writing of Nick Bostrom, and I spent some time trying to come up with an intellectually satisfying account of realist ethics with a techno-utopian slant. I read Josh Greene's thesis - The Terrible, Horrible, No Good, Very Bad Truth About Morality and What To Do About It on moral antirealism and the evolutionary origins of morality, and almost overnight much of my worldview changed to a more humanistic one. I now do voluntary work for the Singularity Institute for Artificial Intelligence.

Artificial Intelligence

The great remaining pre-paradigm science

Looking back over the history of human knowledge, we see that many fundamental questions went from confusing, mysterious unknowns to highly quantitative scientific disciplines with well-established methodologies, conventions and a specialized nomenclature. The mystery of intelligence has yet to undergo this transition - it is, to use Kuhn's terminology, a pre-paradigm science, which means that it is exciting, chaotic and unpredictable.

My research on AI

Representing knowledge about the real world: probabilistic, logical and categorical techniques. Technical report written for the HP-SORCE project as part of the HP Innovation Research Program.

Download PDF [PDF, 2.7MB]

Abstract:

A critical overview of some important formalisms for knowledge representation is presented. This includes knowledge representation using Description Logics(DLs) and First Order Logic (FOL) in the context of both the Semantic Web and artificial intelligence. An overview of the formalism of institution theory is given, along with some background in both category theory and logic, and some motivation in terms of using institutions for structured ontology specification. The hypothesis that institution theory and the HETS system for heterogeneous specification can be applied in the context of ontologies and the Semantic Web is considered, and it is shown that there is evidence suggesting that DLs, FOL and a whole class of formal logics are fundamentally unsuited to representing knowledge about the real world. This indicates that the hypothesis under consideration is likely to be false. Alternative methods for representing knowledge are considered, including probabilistic logics, and it is shown that they may be capable of overcoming the problems identified with model-theoretic logics. A novel proof is given constraining the compatibility of probability theory and institution theory. It is concluded that the Semantic Web could be based upon a semantics that is not well suited to the tasks it intends to carry out, and that it might be useful to attempt to extend techniques such as institution theory so that they can deal with a probabilistic semantics.

AI Safety and the future of the human race

Advanced artificial intelligences with cognitive capabilities significantly in excess of the level of a human being could be used as a weapon of extreme power, and under plausible circumstances could constitute independent actors with goals of their own. This means that they are potentially extremely dangerous, and are probably an existential risk. To find out more about AI and global risks, look at Artificial Intelligence as a Positive and Negative Factor in Global Risk

The idea of humans being wiped out by a smart computer program is usually not taken seriously - it just seems "silly" or "unreasonable". My sentiments about attempting to guess what future technology will do to the world by imposing one's sense of reasonableness as a constraint are captured by this 1939 quote from Winston Churchill:

Atomic energy might be as good as our present-day explosives, but it is unlikely to produce anything very much more dangerous.

Answering the big questions

We are lucky enough to live at a time when life is particularly exciting from a "big questions" point of view. Will the human race survive the twenty-first century and go on to colonize the universe, or will we come this far only to fall at the last hurdle by causing our own extinction? Will scientists unlock the secrets of consciousness and intelligence? Will we develop a fundamental understanding of consciousness and first-person experience? Will we live long enough to live forever?

Human cognitive biases

In the beginning there was evolution, and roughly 200,000 years ago evolution created homo sapiens with the gift of intelligence. But it also plagued us with a whole host of cognitive disabilities and shortcomings; the curse of human cognitive bias. Today we are lucky enough to be able to treat our own mental disabilities by reading the literature on human cognitive bias and compensating.

Transhumanism, personal immortality and the prospect of technologically enabled utopia

We live in a universe that works on physical laws - simple rules that determine how quantum mechanical amplitude evolves over time. The universe is, at its lowest level, rather like a lego kit with a comparatively small number of bricks, or to put it another way, the universe is a lot like Conway's Game of Life. This surprising (and highly counterintuitive) fact has an even more surprising implication: there is no fundamental law that prevents us from improving our lives to arbitrarily high levels of personal well-being. There is no fundamental reason why we cannot create a utopia - to shape the little blocks that the universe is made out of into an arrangement that satisfies our true desires to the maximal extent possible.

Death is not a fundamental fact of existence, it is not a deliberate plan hatched by a careful creator who works in mysterious ways, it is not a punishment or a way to teach us a lesson. It is simply an accident of evolution, and it is perfectly possible (in principle!) to get rid of that particular accident. Likewise human suffering: from the minor annoyances to the agony of losing a loved-one or being betrayed by a friend or partner. Likewise the problems that we humans probably don't even realize that we have, because they are as ubiquitous and invisible to us as water is to a fish. Likewise the extreme suffering of the poorest 1 billion humans.

Transhumanism is simply the idea that we can and should use technology to enable human beings to live the lives that they would, upon reflection, choose. Nick Bostrom put this set of ideas particularly eloquently in his Letter from Utopia:

How can I tell you about Utopia and not leave you nonplussed? What words could convey the wonder? What inflections express our happiness? What points overcome your skepticism? My pen, I fear, is as unequal to the task as if I had tried to use it against a charging elephant.

Have you ever known a moment of bliss? On the rapids of inspiration, maybe, where your hands were guided by a greater force to trace the shapes of truth and beauty? Or perhaps you found such a moment in the ecstasy of love? Or in a glorious success achieved with good friends? Or in splendid conversation on a vine-overhung terrace one star-appointed night? Or perhaps there was a song or a melody that smuggled itself into your heart, setting it alight with kaleidoscopic emotion? Or during worship?

If you have experienced such a moment, experienced the best type of such a moment, then a certain idle but sincere thought may have presented itself to you: “Oh Heaven! I didn’t realize it could feel like this. This is on a different level, so very much more real and worthwhile. Why can’t it be like this always? Why must good times end? I was sleeping; now I am awake.”

Meta-ethical antirealism

Human beings talk about "doing the right thing", as if there were some notion of "rightness" that existed independently of any particular person who might judge things as right or not right. Though this stance - naive metaethical realism - is intuitively compelling, it is false. Understanding the truth about morality (that "the right thing" is whatever we happen to, upon reflection, decide to define it as) is an important prerequisite to improving our lives and to cooperating with other people whose notion of "rightness" is different than our own.

Human extinction and existential risks

The continued existence of the human race is far from guaranteed. There are many possible ways that human beings could be wiped out this century, many of the technologically enabled risks such as genetically engineered super-viruses. The subject of literally saving the entire human race receives little serious academic attention - which is something that must change if we are to maximize our chances of survival.

Category theory

Category theory looks at mathematics from a highly abstracted perspective - to get a flavor of the basics of the subject, Steve Awodey's book is an excellent place to start, as is John Baez's site.

My Part III Essay on operads and the opetopes

The opetopes are a family of higher-dimensional objects that play a role analogous to arrows in ordinary category theory. They provide a possible route to a rigorous theory of weak higher categories, and are incredibly beautiful in themselves. It is a pity that humans can only see in 3 dimensions, because the opetopes would undoubtedly look better in their native higher dimensional worlds. Luckily, my Part III essay explores a 2-dimensional representation of the opetopes using constellations and zoom complexes.

Operads supervised by Prof. Martin Hyland, submitted for the Certificate of Advanced Study in Mathematics

Download PDF [PDF, 2MB]

p-constellation

Media Enquiries

I am available to comment on issues relating to the future development of technology, including:
  • The technological singularity
  • The future of artificial intelligence
  • Existential risks
  • Medium/long term technological risks including nanotechnology, "grey goo", and AI/robotics
  • Transhumanism
  • Cryonics
  • Are we living in a simulation?

Contact me

The best way to contact me is by email; email me at "rmijic" followed by the "at" symbol, followed by "googlemail.com".