Artificial Intelligence

Making sure Artificial Intelligence (AI) does what we want and behaves in predictable ways will be crucial as the technology becomes increasingly ubiquitous. It is an area frequently neglected in the race to develop products, but DeepMind has now outlined its research agenda to tackle the problem.
AI safety, as the field is known, has been gaining prominence in recent years. That is probably at least partly down to the overzealous warnings of a coming AI apocalypse from Elon Musk and Stephen Hawking. It is also recognition of the fact that AI technology is quickly pervading all aspects of our lives, making decisions on everything from what movies we watch to whether we get a mortgage.
That is why DeepMind hired researchers who specialize in foreseeing the unforeseen consequences of the way we built AI back in 2016. The team has spelled out the three key domains they think require research if we are going to build autonomous machines that do what we want.

In a new blog designed to provide updates on the team’s work, they introduce the ideas of specification, robustness, and assurance, which they say will act as the cornerstones of future research. Specification involves making sure AI systems do what their operator intends; robustness means a system can cope with changes to its environment and attempts to throw it off course; and assurance involves our ability to understand what systems are doing and how to control them.

Practical Artificial Intelligence

Researchers tout the potentials of artificial intelligence (AI) as a game changer in a range of industries, but AI appears to have application in the world of gambling as well.

You may not have thought about using artificial intelligence for your Kentucky Derby bets, but those who did, turned their $20 to $11,000. The artificial intelligence, which had earlier predicted the winners of the Super Bowl and the Oscars, made a prediction of the winners in the last recent Kentucky Derby.

The odds for predicting the top four horses in the right order was 540 to one, but this was made possible with swarm intelligence, which amplifies, instead of replaces human intelligence. Swarm uses large groups as they are better at predicting the outcome of an event compared with any one person.

"Research shows that when animals in nature come together in swarms, they can enhance their intelligence to levels they could not have as individuals. UNU asked 20 people who claimed to be knowledgeable about the Kentucky Derby to winnow the horses to the top four and then had the human swarm choose the winning order. The group eventually guessed the winners of the game. Just as the swarm picked, Nyquist took the first place and was followed by Exaggerator, Gun Runner, and Mohaymen. It took about 20 minutes for the AI swarm to pick out the bets. Relying on the swarm's prediction, Unanimous made a $20 bet and won $10,800. Not one in the human group individually predicted the correct order of the horses.

A swarm tends to be more accurate compared with a poll because a poll merely gives the most popular answer and not the answer the optimizes the group's preference.

Moravec's Paradox

Hans Moravec, adjunct faculty member at the Robotics Institute of Carnegie Mellon University, pointed out that machine technology mimicked a savant infant. Machines can do long math equations instantly and beat humans in chess, but they can't answer a simple question or walk up a flight of stairs (until recently). He, along with many others has been working to solve that paradox and help computers evolve on their own.

Early artificial intelligence (AI) researchers believed intelligence was characterized as the things that highly educated scientists found challenging, such as chess, symbolic integration, and solving complicated word algebra problems. They thought, if those could be done so easily by computers, things that children of four or five years could do effortlessly, such as visually distinguishing between a coffee cup and a chair, or walking around on two legs, or responding to words would be infinitely easier for computers to learn.

Computers/robots are finally beginning to move and think like people. Narrative Science can write earnings summaries that are indistinguishable from wire reports. We can ask our phones, 'I'm lost, help.' and our phones can tell us how to get home. (The smartphone was introduced in 2007, just seven years ago.)

Computers that can drive cars were never supposed to happen and ten years ago, many engineers said it was impossible. Navigating a crowded street requires a combination of spacial awareness, soft focus, and constant anticipation. Yet, today we have Google's self-driving cars and they have been approved by some states as allowable on city streets. Ten years from impossible to common.

IBM, working with Memorial Sloan-Kettering cancer information is using its computers to diagnose diseases and the Cleveland Clinic to help train aspiring physicians. It just invested a billion dollars to set up 'Watson' into a separate business unit for medical and other complex decision making activities.

Bottom line, we are experiencing solutions to the paradox and it is very exciting, although I am not sure machines will ever replace the following or that we will ever want to.