If you buy that artificial intelligence is a once-in-a-species disruption, then what Demis Hassabis thinks should be of vital interest to you. Hassabis leads the AI charge for Google, arguably the best-equipped of the companies spending many billions of dollars to bring about that upheaval. He's among those powerful leaders gunning to build artificial general intelligence, the technology that will supposedly have machines do everything humans do, but better. None of his competitors, however, have earned a Nobel Prize and a knighthood for their achievements. Sir Demis is the exception--and he did it all through games.

A natural way to design a negotiation dialogue system is via self-play RL: train an agent that learns to maximize its performance by interacting with a simulated user that has been designed to imitate human-human dialogue data. Although this procedure has been adopted in prior work, we find that it results in a fundamentally flawed system that fails to learn the value of compromise in a negotiation, which can often lead to no agreements (i.e., the partner walking away without a deal), ultimately hurting the model's overall performance. We investigate this observation in the context of the DealOrNoDeal task, a multi-issue negotiation over books, hats, and balls. Grounded in negotiation theory from Economics, we modify the training procedure in two novel ways to design agents with diverse personalities and analyze their performance with human partners. We find that although both techniques show promise, a selfish agent, which maximizes its own performance while also avoiding walkaways, performs superior to other variants by implicitly learning to generate value for both itself and the negotiation partner. We discuss the implications of our findings for what it means to be a successful negotiation dialogue system and how these systems should be designed in the future.

Self-driving cars could soon be able to classify you as a selfish or altruistic driver. While this might bruise some egos, researchers from MIT CSAIL claim that this will make autonomous vehicles (AVs) much safer when driving alongside humans. Predicting how humans might behave, and adjusting an algorithm's reasoning based on how selfish or selfless their behavior might be, could dramatically reduce accidents between AI-enabled vehicles and humans. Properly integrating AI technology with the complicated and nuanced world of human behavior is a huge barrier to overcome, especially in applications that can make a difference between life or death. Apart from making self-driving cars safe enough for our streets, teaching AI how to comprehend the less quantifiable parts of life could give AI the ability to help humans in roles it previously could not handle, and could advance AI applications in general.

Our personalities impact almost everything we do, from the career path we choose to the way we interact with others to how we spend our free time. But what about the way we drive--could personality be used to predict whether a driver will cut someone off, speed, or, say, zoom through a yellow light instead of braking? There must be something to the idea that those of us who are more mild-mannered are likely to drive a little differently than the more assertive among us. At least, that's what a team from MIT's Computer Science and Artificial Intelligence Laboratory (CSAIL) is betting on. "Working with and around humans means figuring out their intentions to better understand their behavior," said graduate student Wilko Schwarting, lead author on the paper published this week in Proceedings of the National Academy of Sciences. "People's tendencies to be collaborative or competitive often spills over into how they behave as drivers.

The rise of artificial intelligence (A.I.) based systems has the potential to benefit adopters and society as a whole. However, these systems may also enclose potential conflicts and unintended consequences. Notably, people will only adopt an A.I. system if it confers them an advantage, at which point non-adopters might push for a strong regulation if that advantage for adopters is at a cost for them. Here we propose a stochastic game theoretical model for these conflicts. We frame our results under the current discussion on ethical A.I. and the conflict between individual and societal gains, the societal value alignment problem. We test the arising equilibria in the adoption of A.I. technology under different norms followed by artificial agents, their ensuing benefits, and the emergent levels of wealth inequality. We show that without any regulation, purely selfish A.I. systems will have the strongest advantage, even when a utilitarian A.I. provides a more significant benefit for the individual and the society. Nevertheless, we show that it is possible to develop human conscious A.I. systems that reach an equilibrium where the gains for the adopters are not at a cost for non-adopters while increasing the overall fitness and lowering inequality. However, as shown, a self-organized adoption of such policies would require external regulation.

In nature, flocking or swarm behavior is observed in many species as it has beneficial properties like reducing the probability of being caught by a predator. In this paper, we propose SELFish (Swarm Emergent Learning Fish), an approach with multiple autonomous agents which can freely move in a continuous space with the objective to avoid being caught by a present predator. The predator has the property that it might get distracted by multiple possible preys in its vicinity. We show that this property in interaction with self-interested agents which are trained with reinforcement learning to solely survive as long as possible leads to flocking behavior similar to Boids, a common simulation for flocking behavior. Furthermore we present interesting insights in the swarming behavior and in the process of agents being caught in our modeled environment.

Self-driving cars are just around the corner, but working out the rules that should govern them is proving tricky. Should they mimic often self-interested human decision making, or be programmed to consider the greater good? It turns out that when you let people program autonomous vehicles themselves, the gap between self-interest and the greater good shrinks. Much of the focus in this area has been on the most pressing moral dilemmas--how should autonomous vehicles behave in cases of life or death? A 2016 study found that people broadly supported utilitarian programming models that save the lives of the most people even if it puts the occupants at risk.

Multi-player Multi-Armed Bandits (MAB) have been extensively studied in the literature, motivated by applications to Cognitive Radio systems. Driven by such applications as well, we motivate the introduction of several levels of feedback for multi-player MAB algorithms. Most existing work assume that sensing information is available to the algorithm. Under this assumption, we improve the state-of-the-art lower bound for the regret of any decentralized algorithms and introduce two algorithms, RandTopM and MCTopM, that are shown to empirically outperform existing algorithms. Moreover, we provide strong theoretical guarantees for these algorithms, including a notion of asymptotic optimality in terms of the number of selections of bad arms. We then introduce a promising heuristic, called Selfish, that can operate without sensing information, which is crucial for emerging applications to Internet of Things networks. We investigate the empirical performance of this algorithm and provide some first theoretical elements for the understanding of its behavior.