The rise of Machine Learning
The terms Machine Learning and Artificial Intelligence (AI) are increasingly common place. AI and Machine Learning make the news, sometime in a positive sense, sometimes not. But they are often portrayed as fledgling technologies, just now finding occasional places in our lives. But nothing could be further from the truth. The machines are here, they will stay here, and they will become ever more invasive.
In essence the terms refer to computers identifying reusable and repeatable patterns, and applying them to problems to ‘intelligently’ find solutions. The temptation is treat these as party pieces, novelties, but doing so would completely miss how integral they have already become in every part of our lives, and how dominant they will become in the future.
While it has been the realms of science fiction for 100s of years, it really became a discipline in 1950, when Alan Turing purported that a machine ‘may learn’. Playing checkers was achieved two years later, and our first party piece was complete.
You may believe progress wasn’t that dramatic. After all, it wasn’t until 2016 that it finally won a game of Go against Fan Hui, (Go being believed to be the most complex game in the world). https://www.wired.com/2016/01/in-a-huge-breakthrough-googles-ai-beats-a-top-player-at-the-game-of-go/
Not so impressive, if you consider at all the things the average 68 year old has learned in the same time frame...
But these novelty headlines are not the real story of AI.
To understand how much of your life is directed by AI consider that it is being increasingly used in healthcare, and clinical trials.
- https://techcrunch.com/2017/03/16/advances-in-ai-and-ml-are-reshaping-healthcare/
- https://www.techemergence.com/machine-learning-in-pharma-medicine/
Governments use AI to divert vast resources for infrastructure, education and health.
- http://www.govtech.com/computing/5-Ways-Artificial-Intelligence-Is-Already-Changing-Government.html
- https://www.allerin.com/blog/machine-learning-to-shape-the-future-of-government
Your pension company and banks use AI to manage your wealth and ensure you and your loved ones have the future you hoped for.
And of course, the military powers of the world use them to protect us and, not to put it too bluntly, kill people who may not be aligned with the same ambitions.
Even the press you read and the court cases you hear about are using AI:
- http://www.digitaljournal.com/business/lawyers-are-turning-to-machine-learning-to-ease-caseloads/article/501622
- https://techcrunch.com/2017/01/28/artificial-intelligence-and-the-law/
- https://www.techemergence.com/automated-journalism-applications/
But at its core it is little more than some matrix operators and a regression algorithm. So how can it have gained such credibility? AI is not a science and not all AI is good or accurate. In many cases, it is flawed and in all the others it is approximate.
- https://www.kdnuggets.com/2015/03/all-machine-learning-models-have-flaws.html
- http://hunch.net/?p=224
But the perception is, if AI’s involved what can go wrong? Of course, difficult papers on limitations of relu, and gradient descent, parameter sensitivity and loss functions are hardly likely to make the news in the same way a machine’s ability to recognize a cat on YouTube does https://www.wired.com/2012/06/google-x-neural-network/ .
There are some high-profile cases of AI failing, https://www.techrepublic.com/article/top-10-ai-failures-of-2016/ , a laughing Alexa https://arstechnica.com/gadgets/2018/03/unprompted-creepy-laughing-from-alexa-is-freaking-out-echo-users/, and many others including self-driving cars and stock market crashes. But hey, whatever… it can find a cat on YouTube… Cool…
AI is early in its development, and yet we allow machines to direct trillions of $ in pensions, government spending, oil drilling, missiles, election efforts, our attention on YouTube, the press, crops, education, and our shopping. It is hard to imagine any area of our lives, corporate or personal that we will not let it influence as it develops.
Perhaps one of the more sobering thoughts is that AI learns from data. We have plenty of that as we continuingly pat ourselves on the back with Big Data headlines. But what learnable behaviors have humans provided for machines? What does that data say about humans co-existing, and what would a true artificial intelligence learn from that data, which it would then enact? Would it learn that humans are industrious and keen to coexist on the earth with other forms of intelligence, or would it learn inequality, injustice, war, politics, survival of the fittest?
And if it took us as its primary source of learning, how would that work out for us?
Just maybe, they have already learned to distract us with cats on YouTube, while they get on with taking over the world?
Interesting to see what can be done in machine learning, nice perspective.