Ari Heljakka, MSc.

Ari Heljakka, MSc. is CTO of Dream Broker Ltd. Dream Broker delivers fast and easy-to-use software for cost-effective video generation by recording your computer desktop view or web-camera feed, along with Internet-based video management and hosting.
 
Ari earned his Masters in Physics and a Bachelors in theoretical philosophy from the University of Helsinki. He has more than a decade’s worth of experience in software development and authored the open-source software project AGISim, as well as the inference engine of the Novamente Cognition Engine. He was recently AI Engineer, Novamente LLC and CEO of GenMind Ltd., a tech startup company funded by the National Tech Agency of Finland, which completed AI research aimed at developing robot cognition software based on the Novamente AI engine.
 
He served as the conference chair of Transvision 2006 and was involved in a panel discussion in May of the same year called How Do We More Greatly Ensure Responsible AGI? at the Artificial General Intelligence Research Institute Workshop. (Transcript is also available.) There he posited the following concern regarding the risks involved in creating human-equivalent computer intelligence: “AGI is potentially extremely capable of carrying out any kind of action on this planet that humans could do right now — and an infinite number of more dangerous and more beneficial actions than we can do at this point. I suppose that’s the premise that we all start from. Then there is the next question — what kind of artificial general intelligence is it going to be possible to create at all? And that’s something we don’t actually know the answer to.
 
“My point, very briefly put, is that we cannot really answer the questions of how to ensure responsible AGI before we have more information about what kind of architectures will it actually be feasible to produce, and what sort of behavior they show in the initial stages. Here, I suppose that we will actually have initial stages, further than we are right now, but not so close, beyond or equal to human-level intelligence as to actually become dangerous.”
 
Ari coauthored Probabilistic Logic Networks: A Comprehensive Framework for Uncertain Inference, Exploring Android Developmental Psychology in a Simulation World, Toward a Pragmatic Understanding of the Cognitive Underpinnings of Symbol Grounding, An Integrative Methodology for Teaching Embodied Non-Linguistic Agents, Applied to Virtual Animals in Second Life, Using Dependency Parsing and Probabilistic Inference to Extract Relationships between Genes, Proteins and Malignancies Implicit Among Multiple Biomedical Research Abstracts, Probabilistic Logic Based Reinforcement Learning of Simple Embodied Behaviors in a 3D Simulation World, and Quantifying Weight of Evidence in Uncertain Inference via Hybridizing Confidence Intervals and Imprecise Probabilities.
 
Watch AGI – What are the Risks?.