Ad blocker interference detected!
Wikia is a free-to-use site that makes money from advertising. We have a modified experience for viewers using ad blockers
Wikia is not accessible if you’ve made further modifications. Remove the custom ad blocker rule(s) and the page will load as expected.
Interview with Shane Legg
by Jonathan Despres. Go to the Interviews.
Tell us about yourself. What is your background, and what current projects are you involved in?
My degrees cover mathematics, statistics, economics, computer science and artificial intelligence.
I've had the good fortune to be able to work with a range of significant people in artificial intelligence and related topics over the last 15 years.
At the moment I'm interested in the effect of cognitive bias in financial markets, algorithms for optimal temporal difference learning, advanced types of artificial neural network architectures for high dimensional temporal prediction and control, and also issues surrounding the safety of advanced artificial intelligence.
What are your goals for the next decade?
To follow my dreams and lead an interesting and enjoyable life.
In the last 5 years we have really started to make good progress on getting recurrent artificial neural networks to perform. I think this is an area ripe for a revolution, and I hope to be a part of it.
I am interested in the application of machine learning to problems in finance, so I might well find myself involved with that.
I would like to make a significant contribution to AI safety, however this is a very tricky problem so unfortunately I'm not expecting much.
When do you think will we achieve real life extension?
I don't know. Perhaps by 2035?
Your vision of the future?
AI is now where the internet was in 1988. Demand for machine learning skills is quite strong in specialist applications (search companies like Google, hedge funds and bio-informatics) and is growing every year. I expect this to become noticeable in the mainstream around the middle of the next decade. I expect a boom in AI around 2020 followed by a decade of rapid progress, possibly after a market correction. Human level AI will be passed in the mid 2020's, though many people won't accept that this has happened. After this point the risks associated with advanced AI will start to become practically important.
Do you believe in Cryonics and when will it suceed?
What kind of jobs you did when you were younger and what is the important things you learned about it?
I worked for 2.5 years as a database programmer after my Master's degree. I learned that a regular job in a regular office doing regular things was totally and absolutely not for me.
Why isn't the science of cryonics progressing at a rate commensurate to other sciences?
I don't know.
Do you believe in a God?
No. Or more accurately, I think it's unlikely but not completely impossible.
What extropian values do you prefer the most and why?
Don't know much about extropians.
What would you love to accomplish before you die?
Invent a powerful AGI and have it make the world a much better place.
What are the languages of the universe for you?
What do you think about the singularity, when will it happen?
I don't know about a "singularity", but I do expect things to get really crazy at some point after human level AGI has been created. That is, some time from 2025 to 2040.
When do you think AI Robots will surpass humans?
Machines can already surpass humans, and all other animals, in countless ways.
What do you think of the movie AI by Steven Spielberg?
If the "boy" had died when he hit the water it would have been an ok film. Instead we had to endure a ridiculous sugar frosted Hollywood ending.
Will Sexbot be possible and when?
I have no interest in such things as I don't think they will be of any consequence compared to super intelligent machines.
What would be a super intelligent machine ?
A machine that can achieve a wide range of goals in a wide range of environments. For a more exact definition, indeed a formal mathematical definition, see my thesis:
Why would it be useful ?
The same reason that intelligence is useful.
Your best movie ever is?
Your religion is?
Your political view?
Half way between a libertarian and a centreist.