Ari Durresi

Professor of Computer Science at Indiana University Indianapolis. Interested in Trust Engineering, Trustworthy and Causal AI, AI Control, Security, Networks

[Ari Durresi @AriDurresi]

@HowToPrompt__ Einstein knew about Mercury's orbit that Newton's mechanics could not explain.

[Ari Durresi @AriDurresi]

@drmichaellevin This is a great idea. Other fields should embrace it.

[Ari Durresi @AriDurresi]

@ChrisPalmerMD Great results and hope for people with these conditions. Causal AI can help us go beyond correlations and discover underlying mechanisms.

[Ari Durresi @AriDurresi]

Consciousness has emerged during evolution to serve some purpose for living organisms. We can model kinds of consciousness in AI, such as world models, to better serve our purposes. But if it is possible, we should be very careful about AI's real consciousness; we might not like its self-serving purposes.

[Ari Durresi @AriDurresi]

@yudapearl @CeesvanderVelde @daveg Probably they meant ANI. N-Non

[Ari Durresi @AriDurresi]

@Kasparov63 The idea of intention is a rabbit hole. So an own goal should not count.

[Ari Durresi @AriDurresi]

@yudapearl @CSProfKGD @AndreasBikfalvi The world is causal, so good models of it should be as well. All those disciplines, such as physics, etc., which are partial models of reality, have one thing in common: causality.

[Ari Durresi @AriDurresi]

@yudapearl @UCLAengineering @eliasbareinboim @GaryMarcus @GadSaad Congratulations, Prof. Pearl!

[Ari Durresi @AriDurresi]

@Kasparov63 Probably our best shot against mean aliens. We can convince them to try it…

[Ari Durresi @AriDurresi]

@BernardJBaars Metaphors highlight where the attention is directed. At some level, it is similar to consciousness.

[Ari Durresi @AriDurresi]

@NIHDirector_Jay Causal AI can power clinical trials to discover mechanisms underlying health conditions.

[Ari Durresi @AriDurresi]

@yudapearl @KirkDBorne If AI can justify the I in it, it has to be causal.

[Ari Durresi @AriDurresi]

@ChrisPalmerMD Metabolism affects all aspects of health. It would have been very strange not to affect mental health. Finding the causal paths will help the field and many people. Causal AI is here to lend a hand.

[Ari Durresi @AriDurresi]

@ChrisPalmerMD @QuietBiology This is a good target for causal AI to sort out practical causality in such complex relations.

[Ari Durresi @AriDurresi]

@newstart_2024 Fasting is probably the closest thing to a health miracle. It's the way all animals survive in natural conditions.

[Ari Durresi @AriDurresi]

@sirbayes @YujiaZheng9 Modeling the full world might be impossible, as shown by the status of Theories of Everything. But we have theories of various aspects of reality. So AI task-specific models are the way to go.

[Ari Durresi @AriDurresi]

@drfeifei @a16z “All models are wrong; some of them are useful.”
So the goal is to build the most useful one for the given application.

[Ari Durresi @AriDurresi]

@MushtaqBilalPhD They are cheating themselves.

[Ari Durresi @AriDurresi]

@ChrisPalmerMD An amazing angle of such intervention is its scalability. Many people could benefit from these results. That’s what science is about.

[Ari Durresi @AriDurresi]

@adamsafron @yudapearl @eliasbareinboim @GaryMarcus @DavidDeutschOxf @thesmartrobot Thanks for the reference. Very interesting. It is great to get inspired and use all the mechanisms that evolution has developed. On the other hand, we can go further, unconstrained by the constraints of biological systems.