Transcript PowerPoint-esitys

HOW TO MAKE
A MACHINE
THINK?
CONTACT DETAILS
PITCH OVERVIEW – NOT
AN ACTUAL SLIDE
• Our team constists of 7 undergraduate students fascinated by
consciousness and AI.
• We are aiming to simulate human-like thinking on a computer.
• This new AI is based on a semantic memory system of our own design.
• Contrary to we are doing, current AI applications concentrate only on
modelling a single narrow subfield of human cognition at a time.
CHALLENGE
• How, if at all, is it possible to implement human-like thinking
into a computer?
• New kind of artificial intelligence could help explaining consciousness
THE TEAM
• 7 undergraduate students fascinated by consciousness and AI
• Team strengths:
+ impressive amount of new ideas
+ knowledge in various fields of study
+ self-reflection, problem solving and creativity to solve the impossible
+ programming and math skills
+ desire to aim high
BACKGROUND TO THE
APPROACH
• The traditional artificial intelligence viewpoint:
Cognition/The brain consists of independently working separate
modules.
• Despite decades of research no human-like AI system has been
demonstrated.
• We believe such a system can be build by using a different approach.
SOLUTION
• Our approach:
Human-like thinking requires semantic, tightly interconnected cognitive
mechanisms.
• What is innovative in our model of human cognition is that includes
 semantic memory
 semantic problem solver
 language module
 conscious reasoning
• The model is also mechanical:
 no unexplainable steps
 totally programmable
THE IMPACT
• If we succeed, we will have
An artificial intelligence that can
 adapt to its surroundings
 solve problems creatively
 process meanings in the same way as humans do
• An unprecedented number of applications become feasible
• Maybe even a human-like robot in the far future!
CALL TO ACTION
• We wish opportunities to discuss and refine details of our model with
academic experts on various fields.
• Also visibility and contacts with companies interested in our work would
be most welcome.