Saturday, March 12, 2016

A few thoughts on creating general purpose artificial intelligence

I recently gave a presentation describing various methods used to accomplish machine learning techniques used for bipedal walking. The most successful algorithms typically use multiple artificial intelligence techniques in order to accomplish this. My presentation focused on using a combination of neural networks and genetic algorithms or harmony search algorithms. The configuration for the neural network is set up as follows. The neural network is configured such that the input layer receives sensor values from various points on the robot such as touch sensors on the feet and encoders to get the angle of joints. The data is then sent through the layers of the neural network and it is used to produce an output value that controls where the legs are moved. The output values and are then sent back into the neural network and used to update the weights on each node to improve future performance. The addition of a genetic algorithm or harmony search algorithm makes it possible to produce more accurate weights. The challenge that is presented is the fact that this creates a neural network that is highly specialized in accomplishing walking. This network could be sufficiently advanced enough to handle uneven surfaces walking on incline planes however, there is still a significant challenge because this network could be performing excellently in all scenarios except if a ladder is suddenly encountered for at for example, then it may not be able to find a solution that works correctly.

The concept of the general artificial intelligence, which would otherwise be described as a machine that has all of the thinking and computational abilities of human being, is someone of a challenge to developers. Looking deeper at the function of the human brain, it becomes clear that the human brain is not actually ideally suited for all scenarios. As a matter of fact, the human brain is divided up into a plethora of highly specialized areas. I think it may then be reasonable to assume that the same solution can be accomplished in artificial intelligence. The utilization of multiple neural networks where each becomes specialized to accomplish a specific task could be a solution to this problem. A primary neural network could be used to map input values to the correct sub-neural network. In addition it may be useful to look into the potential ability for this network to dynamically create new sub-neural networks that could be automatically specialized to accomplish some kind of new task.

Saturday, February 20, 2016

Rise of the Robots: Technology and the Future of Work

I went to see a the keynote address of Rise of the Robots. Technology and the Future of Work presented by Martin Ford at Mount Holyoke College this weekend. Ford is an entrepreneur and author who has brought to light a selection of very interesting trends in employments and their relation to automation.

For nearly any employee in almost every occupation there is always the possibility that a new technology will automate some task that their position previously required a human to achieve. We have now reached the point where there are very few tasks that humans can accomplish that machines cannot. In his presentation, Ford points out that the perceived threat of automation to the workforce is nothing new. Automation has been replacing jobs since the turn of the century. Industrialization in the United States replaced countless jobs that were previously done by hand by factory workers.

As it currently stands, it is not a question off whether robots will replace blue collar jobs in the future. All occupations will be effected in some way. Ford points out that in an ideal scenario, an employee would be re-trained for a higher level position once their job was replaced by an automated process. Unfortunately, this is not the case within many companies. In the past there was a direct correlation between an increase in employee productivity (as a result of technological improvements) and the pay that the employee received. This makes a lot of sense because it only seems reasonable that as a productivity increases, the company's profits increase. As a result, employees should see a returned increase of their share of that success. This was true until 2006, when the average amount of payment that employees receive in proportion to the productivity of employees stopped following that trend. While employee productivity continues to increase, wages have remained relatively unchanged.

The real threat to employment does not appear to be automation itself, but automation that makes employees less valuable. In the past, an individual who was trained to use a particular tool or technology was valued by a company and could expect a higher wage because of their skill or knowledge. This is no longer the case because as computers get more advanced, the process of automating task requires little or no human supervision or insight.

The explosion of advances in automation and artificial intelligence have created some difficulties for companies. Why should a corporation employ a human worker rather than a robot that will never get tired or hurt and can perform in a superior manner to their human counterpart? In the future, it appears that the wide availability of robotic technologies will make it such that it it will not be profitable for companies to employ workers.

It is possible that increased advances in technology will create new job opportunities within fields that are impossible to predict at the current time. Just as a job as a social media marketer could never have been anticipated as a possible career several years ago, it is possible that new research into synthetic biology or nanotechnology will yield new occupations for the next generation of workers.

At this point in the discussion, one begins to wonder if there will be any limitation to what can possible be accomplished with technology. It is not unlikely that the computer architects of the future will be other computers, creating new systems that are far more efficient than themselves which will then, it turn, go on to repeat the process.

Ford concluded with some very interesting suggestions about how we can possibly overcome the possible perils of total automation. He suggests the concept of decoupling jobs from income. He stresses the importance of consumerism that is required for an economy to function. Machines do not consume, so a guaranteed minimum income could be one possible solution to help prevent inflation at the hands of an automated economy.