Some of mankind’s most complex games, like Go, have failed to challenge AIs from the likes of DeepMind. Reinforcement learning trains algorithms by running scenarios repeatedly with a ‘reward’ given for successes, often a score increase.

Two classic games from the 80s – Montezuma’s Revenge and Pitfall! – have thus far been immune to a traditional reinforcement learning approach. This is because they have little in the way of notable rewards until later in the games.

Applying traditional reinforcement learning typically results in a failure to progress out the first room in Montezuma’s Revenge, while in Pitfall! it fails completely.

One way researchers have attempted to provide the necessary rewards to incentivise the AI is by adding them in for exploration, what’s called ‘intrinsic motivation’. However, this approach has shortcomings.

“We hypothesize that a major weakness of current intrinsic motivation algorithms is detachment,” wrote Uber’s researchers. “Wherein the algorithms forget about promising areas they have visited, meaning they do not return to them to see if they lead to new states.”

Uber’s AI research team in San Francisco developed a new type of reinforcement learning to overcome the challenge.

The researchers call their approach ‘Go-Explore’ whereby the AI will return to a previous task or area to assess whether it yields a better result. Supplementing with human knowledge to guide it towards notable areas sped up its progress dramatically.

If nothing else, the research provides some comfort us feeble humans are not yet fully redundant and the best results will be attained by working hand-in-binary with our virtual overlords.

 AI & >.

The post Uber’s AI beats troublesome games with new type of reinforcement learning appeared first on AI News.

Andrew Moore, the new head of AI at Google Cloud, is one such expert who has warned of scenarios that would lead to public backlash and restrictions that would prevent AI from reaching its full potential.

Back in November, Moore spoke at the Artificial Intelligence and Global Security Initiative. In his keynote, he said:

“If an AI disaster happens – and that would, for instance, be an autonomous car killing people due to serious bugs – then at some point AI systems are going to be locked down for development, at least in the US.

There are some even more horrible scenarios — which I don’t want to talk about on the stage, which we’re really worried about — that will cause the complete lockdown of robotics research.”

Autonomous vehicles have indeed already been involved with accidents.

Back in March, just four months after Moore’s warning, an Uber self-driving vehicle caused a fatality. The subsequent investigation found Elaine Herzberg and her bicycle were acknowledged by the car’s sensors but then flagged as a ‘false positive’ and dismissed.

Following years of sci-fi movies featuring out-of-control AI robots, it’s unsurprising the public are on edge about the pace of recent developments. There’s a lot of responsibility on researchers to conduct their work safely and ethically.

Professor Jim al-Khalili, the incoming president of the British Science Association, told the Financial Times:

“It is quite staggering to consider that until a few years ago AI was not taken seriously, even by AI researchers.

We are now seeing an unprecedented level of interest, investment and technological progress in the field, which many people, including myself, feel is happening too fast.”

In the race between world powers to become AI leaders, many fear it will lead to rushed and dangerous results. This is of particular concern with regards to AI militarisation.

Many researchers believe AI should not be used for military purposes. Several Google employees recently left the company over its contract with the Pentagon to develop recognition software for its drones.

Over 4,000 other employees signed a petition demanding that Google’s management cease the project and promise to never again ‘build warfare technology.’

Google has since made the decision not to renew its Pentagon contract when it expires. However, it’s already caused ripples across Silicon Valley with many employees for companies such as Microsoft and Amazon demanding not to be involved with military contracts.

Much like the development of nuclear weapons, however, AI being developed for military purposes seems inevitable and there will always be players willing to step in. Last month, AI News reported Booz Allen secured an $885 million Pentagon AI contract.

From a military standpoint, maintaining similar capabilities as a potential adversary is necessary. Back in July, China announced plans to upgrade its naval power with unmanned AI submarines that provide an edge over the fleets of their global counterparts.

Russian President Vladimir Putin, meanwhile, recently said: “[AI] comes with colossal opportunities, but also threats that are difficult to predict. Whoever becomes the leader in this sphere will become the ruler of the world.”

Few dispute that AI will have a huge impact on the world, but the debate rages on about whether it will be primarily good or bad. Beyond the potential dangers of rogue AIs, there’s also the argument over the impact on jobs.

Al-Khalili wants to see AI added to school curriculums – as well as public information programmes launched – to educate good practices, prepare the workforce, and reduce fears created by sci-fi.

What are your thoughts on AI fears?

The post Experts warn of AI disasters leading to research lockdown appeared first on AI News.

The company is using the data for a new feature called ‘Profile Recommendations’ whereby the app will recommend switching to a correct profile for your journey.

Many people will have two Uber profile – one for personal use, and the other for business.

Ronnie Gurion, GM and Global Head of Uber for Business, says:

“Using machine learning, Uber can predict which profile and corresponding payment method an employee should be using, and make the appropriate recommendation.”

When quickly booking a ride, it can be easy to forget to switch. Accidentally booking a ride home from a night out using a business account set up with a workplace’s payment details is an unwanted conversation with an employer.

Uber believes its success rate for determining the correct profile is around 80 percent.

To help reduce the 20 percent it gets wrong, businesses can assign ‘trip reviewers’ who know whether an employee’s use is supposed to be personal.

Any questionable rides can be flagged by the reviewer and the employee can decide in the app if it was supposed to be a personal ride or not. The whole process is designed to be much quicker than starting email threads about the issue and similar bureaucratic processes.

What are your thoughts on Uber’s use of AI for its latest feature?

The post Uber is using AI to determine if a ride is business or pleasure appeared first on AI News.

The sobriety of a passenger is a big factor in the safety of both driver and rider. Drivers have been assaulted by inebriated passengers, and most of the sexual assaults conducted against Uber riders by drivers have been while they were intoxicated.

According to a CNN investigation, at least 103 Uber drivers have been accused of sexually assaulting or abusing passengers in just the past four years.

Cutting off the service for anyone who’s the slightest bit intoxicated doesn’t make business sense and is likely to cause more problems on a societal level than it solves.

Many use Uber to get home after a few drinks without any problem, and I’m sure most would agree that it’s much better for everyone than if they were to reach for their own car keys.

To determine the user’s level of sobriety, the patent describes Uber’s AI learning how each user typically uses their app. Unusual factors such as inaccurate pressing of buttons, typos, walking speed, and more will be taken into account.

Someone who is determined too intoxicated may be denied a ride, or paired up with a driver with skills or training for dealing with inebriated passengers. For their trouble, these drivers could be able to charge elevated rates.

If nothing else, it may save a few Uber cars from smelling of vomit.

What are your thoughts on Uber’s algorithm for detecting intoxication?

The post Uber wants to develop an intoxication-detecting algorithm appeared first on AI News.