Some research groups
Element AI, Montreal, co-founder prof. Benjo
Deep Mind, London, acquired by Google
Machine learning is the ability to learn and make predictions using data. Data are the input and the output. A diverse set of skills set are required for manipulating large amount of data, defining the problem and the outcome, an understanding of the algorithms and an understanding of a lot of maths and statistics. Overall, machine learning is a complex task that requires advanced knowledge of different and complex subjects.
Fortunately a set of services are available that abstract the level of competencies required and they support data scientists and analysts in their tasks. At the end, what we really want is to make sense of our data and be able to make predictions about them.
The level of abstraction is important and allows the creation of more advanced services on top of machine learning cloud platforms.
Below a list of some of the most well known services in this space:
Google Cloud Machine Learning
Big data and predictive models are currently used and are going to be pervasive affecting our lives. There are many positive news coming from these new algorithms and techniques. At the same time, it is important to be aware of their impact, biases and how to deal with the imperfections and abuses that inevitable will come with them.
Weapons of Math Destruction by Cathy O’Neil is an extremely good and important book on this subject and I highly recommend it to all the people that want to understand how society is and will be affected by all these new algorithms and predictive models.
The Dentist Office Software Story is an excellent blog post by Fred Wilson at avc.com.
There were a couple of sentences what caught my attention and I want to emphasize here. Below the excerpt where I have highlighted the sentences in bold.
This story is designed to illustrate the fact that software alone is a commodity. There is nothing stopping anyone from copying the feature set, making it better, cheaper, and faster. And they will do that. This is the reality that Brad and I stared at in 2003 as we were developing our initial investment thesis for USV. We saw the cloud coming but did not want to invest in commodity software delivered in the cloud. So we asked ourselves, “what will provide defensibility” and the answer we came to was networks of users, transactions, or data inside the software. We felt that if an entrepreneur could include something other than features and functions in their software, something that was not a commodity, then their software would be more defensible. That led us to social media, to Delicious, Tumblr, and Twitter. And marketplaces like Etsy, Lending Club, and Kickstarter. And enterprise oriented networks like Workmarket, C2FO, and SiftScience. We have not perfectly executed our investment strategy by any means. We’ve missed a lot of amazing networks. And we’ve invested in things that weren’t even close to networks. But all of that said, our thesis has delivered for us and we stick to it as much as we can.
In the concluding chapter he writes:
Yet if we take the really grand view of life, all other problems and developments are overshadowed by three interlinked processes:
1. Science is converging on an all-encompassing dogma, which says that organisms are algorithms, and life is data processing.
2. Intelligence is decoupling from consciousness.
3. Non-conscious but highly intelligent algorithms may soon know us better than we know ourselves.
These three processes raise three key questions, which I hope will stick in your mind long after you have finished this book:
1. Are organisms really just algorithms, and is life really just data processing?
2. What’s more valuable – intelligence or consciousness?
3. What will happen to society, politics and daily life when non-conscious but highly intelligent algorithms know us better than we know ourselves?
I have been reading The Inevitable by Kevin Kelly. A great book that I highly recommend.
Here I report a few excerpts from the “Flowing” chapter to describe how flows of data are inevitable and what we can do about it.
The Internet is the world’s largest copy machine. At its most fundamental level this machine copies every action, every character, every thought we make while we ride upon it.
If something can be copied – a song,a movie, a book – and it touches the internet, it will copied.
The information age is driven by digital copies – exact and free.
In this new supersaturated digital universe of infinite free digital duplication, copies are so ubiquitous, so cheap – free, in fact – that the only things truly valuable are those that cannot be copied.
When copies are superabundant, they become worthless. Instead, stuff that can’t be copied become scarce and valuable.
Here are eight generatives that are “better than free”:
Immediacy: Getting something the moment it is released, or even better, produced by its creators.
Personalization: providing something personalized according to your taste, your reading taste, your living room, etc.
Interpretation: interpretation of genomic info, healthcare, travels, consulting Linux, etc.
Authenticity: make sure you have the real thing, the real software app
Accessibility: example are cloud services to access info wherever and whenever you want with full backup, security.
Embodiment: live concerts, live lessons, printed book, Ted talks
Patronage: fans want to pay creators
Discoverability: providing guides or ways to help discover new things, reading lists, videos, movies
Some very good books and debates on growth, innovation and what it means for us.
Robert Gordon, The End of Innovation, the End of Growth
Erik Brynjolfsson, The Key to Growth, Race with the Machines
The future of work and innovation: Robert Gordon and Erik Brynjolfsson debate at TED2013
The Second Machine Age: Work, Progress, and Prosperity in a Time of Brilliant Technologies by Erik Brynjolfsson, Andrew McAfee
World after capital, Albert Wenger
Machine learning algorithms are becoming more “intelligent” and replacing people in many tasks. What’s next?
When (not if) Machine learning will create a superintelligence, then how we will deal with it? How do we control it? How do we avoid to become extinct because of that? Read Nick Bostrom’s book, Superintelligence: Paths, Dangers, Strategies
Is there a unifying algorithm in Machine Learning? Read Pedro Domingos’ book, The Master Algorithm: How the Quest for the Ultimate Learning Machine Will Remake Our World
And while algorithms are increasingly used, they have bias and problems that must be taken into account, see Artificial Intelligence’s White Guy Problem