We study several classes of interactive assistants from the points of view of decision theory and computational complexity. We first introduce a class of POMDPs called hidden-goal MDPs (HGMDPs), which formalize the problem of interactively assisting an agent whose goal is hidden and whose actions are observable. In spite of its restricted nature, we show that optimal action selection in finite horizon HGMDPs is PSPACE-complete even in domains with deterministic dynamics. We then introduce a more restricted model called helper action MDPs (HAMDPs), where the assistant's action is accepted by the agent when it is helpful, and can be easily ignored by the agent otherwise. We show classes of HAMDPs that are complete for PSPACE and NP along with a polynomial time class. Furthermore, we show that for general HAMDPs a simple myopic policy achieves a regret, compared to an omniscient assistant, that is bounded by the entropy of the initial goal distribution.

We study several classes of interactive assistants from the points of view of decision theory and computational complexity. We first introduce a class of POMDPs called hidden-goal MDPs (HGMDPs), which formalize the problem of interactively assisting an agent whose goal is hidden and whose actions are observable. In spite of its restricted nature, we show that optimal action selection in finite horizon HGMDPs is PSPACE-complete even in domains with deterministic dynamics. We then introduce a more restricted model called helper action MDPs (HAMDPs), where the assistant's action is accepted by the agent when it is helpful, and can be easily ignored by the agent otherwise. We show classes of HAMDPs that are complete for PSPACE and NP along with a polynomial time class.

Summary: Contextually intelligent, NLP-based interactive assistants are one of the next big things for AI/ML. The tech is already here from recommendation engines. The need to be more efficient and to become AI-augmented in our decision making is now. Getting the contextual awareness is the hard part. Last week we took the position that from a technical standpoint, 'deeply inclusive and contextually sensitive' AI is one of the two'next big things' in AI.

The Lighthouse interactive assistant monitors the comings and goings at a home remotely, but with an AI touch. SAN FRANCISCO -- Lighthouse, an artificial intelligence services start-up with roots in the self-driving car world, wants to do for the home what the DARPA Grand Challenge did for autonomous cars. The 2-year-old start-up has developed an interactive assistant for the home that, essentially, does the opposite of Alexa, Amazon's voice-activated personal assistant. While Alexa keeps consumers connected to the outside world while they're at home, Lighthouse keeps consumers connected to their homes while they're far away. The interactive assistant, in the form of a mini-lighthouse, leverages deep learning and 3-D sensing technology developed as part of the DARPA Grand Challenge, the world-renowned competition in which autonomous vehicles navigate an off-road course.