I spend a disproportionate amount of time reading and talking to two somewhat niche groups of people in American politics: democratic socialists of the Sen. Bernie Sanders variety (or maybe a bit to the left of that), and left-libertarians from the Bay Area who are interested in effective altruism.

These are both small groups, but they have social and intellectual influence bigger than their numbers. And while from a distance they look similar (I’m sure they both vote for Democrats in general elections, say), there’s a big issue on which they part ways where collaboration could be productive: artificial intelligence safety.

Effective altruists have, for complex sociological reasons I explored in a podcast episode, become very interested in AI as a potential “existential risk”: a force that could, in extreme circumstances, wipe out humanity, just as nuclear war or asteroid strikes could.

Kelsey Piper has a comprehensive Vox explainer of these arguments, and I take them seriously, but most friends to my left do not. A typical reaction is that of Elmo Keep, who dismisses AI doomsday arguments as “the stuff of stoned freshmen who’ve read too much Neal Stephenson.”

It doesn’t help that alt-right funder extraordinaire Peter Thiel has long supported research into AI safety, especially at the Machine Intelligence Research Institute, whose founder Eliezer Yudkowsky is nearly as polarizing as Thiel is.

But you don’t have to go full Silicon Valley libertarian to become convinced that the rising power of AI is a pressing social problem. There are distinctively leftist reasons to be worried.

The most obvious is that the best-funded developers of AI, at least in the US and the UK, are private companies. Insofar as the government’s involved in R&D, it’s largely through the Defense Department.

So: do we want DeepMind, a Google sister company, to be developing sophisticated wargaming technology with minimal regulation? Or do we want to democratically weigh how DeepMind should proceed and make regulations as a society?

But the leftist case for worry is broader than this. AI safety experts often characterize the problem as one of “alignment”: Is the goal for which AI is optimizing aligned with the goals of humanity as a whole? To use an oft-repeated thought experiment, would a sufficiently powerful AI, told to produce (say) paperclips, end up hoarding resources until it destroys the world and uses all the Earth’s resources to build paperclips?

That’s a ridiculous example and one that I find turns off a lot of the not-already-converted, including some eminent AI researchers who find talk of apocalypse overblown. But here’s a less ridiculous one.

Algorithms are already misaligned with human ethics, and it’s already a problem

Let’s say that Pittsburgh wants to more effectively respond to cases of child neglect, and better triage its case workers so they investigate the worst cases. So the city builds a predictive model based on thousands of previous cases that can assign a score to each case that comes through that helps prioritize it.

But using previous cases to train the AI means it winds up baking in a lot of prejudices from previous generations of case workers, who may have been biased against poor and black parents.

This method is not very precise and produces a lot of false positives, but its air of precision and science means that case workers have started to defer to the algorithm. It has gained power in spite of its weak empirical underpinnings.

This is not a hypothetical example. This is the story of the Allegheny Family Screening Tool (AFST), an algorithm used by the county encompassing Pittsburgh to triage child abuse cases. Virginia Eubanks, a political scientist at the University of Albany, chronicles this system and its many limitations in Automating Inequality(2018), the best single book on technology and government I’ve read.

Often cases like this are conceptualized through the prism of rights to privacy, or rights to oversight over algorithms; my colleague Sigal Samuel has done incredible work on this front. But fundamentally it’s an AI alignment problem of exactly the same kind, if not the same scale, that MIRI and effective altruists worry about.

Some in the artificial intelligence research community having been urging researchers and advocates to make these connections. “Some critics have argued that long-term concerns about artificial general intelligence (AGI), or superintelligence, are too hypothetical and (in theory) too far removed from current technology for meaningful progress to be made researching them now,” Cambridge’s Stephen Cave and Seán S. ÓhÉigeartaigh wrote in Nature: Machine Intelligence earlier this year. “However, a number of recent papers have illustrated not only that there is much fruitful work to be done on the fundamental behaviors and limits of today’s machine learning systems, but also that these insights could have analogues to concerns raised about future AGI systems.”

There is a left/libertarian alliance to be made in translating Bay Area worries about the long-run capabilities of AI into worries about the way it’s being deployed right this minute. We could be using these more primitive technologies to build a precedent for how to handle alignment problems that will become vastly more complex with time.

As the Silicon Valley libertarians set their sights to more specific cases, leftists should set them a bit wider. The automation of governance will not stop with Pittsburgh’s family services system. It will be vastly more dramatic in 10, 20, 30 years.

And the only way to ensure alignment is to stop treating this as a silly issue, and actively engage with AI safety.

The post Why the left should worry more about AI appeared first on Artificial Intelligence.

Every company is in the process of understanding how artificial intelligence (AI) will affect their industry. It can be intimidating and even discomforting to consider the consequences of using machines to make decisions. But if there’s an industry that should be comfortable with this potential shift, it’s the marketing industry.

Making decisions based on data isn’t a new concept. It’s long been at the core of any successful marketing strategy or campaign, including at my own agency. So how can digital agencies use their data literacy to take full advantage of AI? And perhaps more importantly, how can digital agencies use AI to improve agency culture and even inspire innovation?

Digital Agencies Are Already Using AI

Before we dive in, let’s first specify what we mean when we talk about AI. In the context of digital marketing, AI is closely tied to machine learning, where computer systems are capable of learning and improving performance through data analysis without human intervention.

One example of AI already in action can be seen in programmatic media buying using a demand-side platform (DSP) such as Google’s DoubleClick Bid Manager. The programmatic buying platform incorporates a variety of AI features, including automated targeting using real-time bidding models, a simplified buying process, automated budget pacing, and real-time reporting and optimization toward the key performance indicator a campaign manager selects. AI can also be used in serving dynamic creative, as algorithms begin to learn which audiences respond to various creative versions or featured products.

Another example of digital agencies integrating AI into their operations is the use of ad rotation settings in Google AdWords. When using the “optimize” setting, the machine learning technology prioritizes search ads that are statistically more likely to perform more efficiently based on keywords, search term, device and location, among other variables.

Tools like Google Analytics are great for collecting data, but similar to DoubleClick, the real value in AI is its ability to analyze that data and endorse strategic action. PaveAI, for example, is doing more than simply communicating information through graphs and charts. It’s using statistical models to recommend actions that are focused on generating leads and sales as opposed to site traffic.

Facebook has also been a major proponent of AI, with a global team (Facebook AI Research, or “FAIR”) dedicated to helping communities further understand how automated systems and processes can achieve human-level intelligence. Despite the fact that Facebook has recently severed ties with many third-party data providers as a result of its public shakeup, AI-influenced advertising campaigns (“smart ads”) on the social media giant’s pages are still impactful when we consider the wealth of targeting information users explicitly make available to Facebook.

It’s important to note that many applications of AI are still in their infancy, and that’s especially true in the content marketing arena. While organizations like the Associated Press are using algorithms to produce basic content like stock reports, we’re likely still a long way from replacing the entire creative process. But many agencies are exploring how AI can help their content marketing, and some are already delivering personalized content at scale based on collected data and analysis.

AI And A New Agency Culture

Discussing the implications of AI can certainly raise a few eyebrows in the workplace. There are plenty of misconceptions surrounding AI, and perhaps the most prominent misunderstanding is that it will lead to employee layoffs. While not every position is immune to these concerns, the primary objective of using AI for digital marketing is to automate rudimentary tasks like data collection, processing and reporting. From my perspective, the idea is that automating these tasks doesn’t lead to layoffs; it leads to more efficient processes and more time for people to focus on strategic planning.

Let’s consider an employee who spends 16 hours a week processing and reporting data that is collected through common marketing tools. This employee first must process the data, which is simply understanding the information that has been gathered. The employee then analyzes the data, looking for trends, inaccuracies or incongruities. The employee draws conclusions and gives recommendations based on their experience and analysis of the data. This process requires extensive experience and an in-depth understanding of statistics, cause and effect, and historical trends.

If this process was automated through AI and/or machine learning, the agency could benefit from the same accurate and reliable data analysis provided by the employee. While the employee was hired in part for their data literacy, it’s more likely they were hired for their marketing expertise, ability to reason, creativity and business instincts.

In my opinion, this is where the agency stands to gain the most from effective AI integration. With 16 hours a week now available, the employee is free to focus on the critical thinking that leads to improved client campaign performance, experiment with new approaches and build meaningful relationships with colleagues and clients. This concept extends to every branch of the agency, fostering a culture where executives, human resources, management and employees are leading and innovating for the benefit of the agency and its clients.

Digital agencies are in a unique position because of their experience and expertise in understanding and using data to make business decisions. Based on what I have seen, many agencies have already been using AI and machine learning to optimize ads and personalize the customer experience — with positive results. This experience allows those at the forefront to quickly adapt to the effects of AI on their industry.

While AI can be a worrisome topic, it actually presents an amazing opportunity for agencies to help their employees, not replace them. Automating simple tasks gives employees time to focus on the primary business objectives of their clients in order to help them succeed. When this theory is applied across the entire agency, I believe a culture of innovation and progress can take hold — pushing the agency, its clients and the entire industry forward in a way that wasn’t possible before.

The post The Impact Of Artificial Intelligence On The Future Of The Digital Agency appeared first on Artificial Intelligence.

Artificial intelligence and machine learning are in buzz these days and more and more people are interested to learn about it. It got a major breakthrough when Google made AIhistory by creating an algorithm that mastered Go. And the technological advancement is creating more jobs as companies need high-skilled AI talents to develop and maintain a wide range of applications.

If you are interested in becoming a machine learning expert but don’t know where to start from? Don’t worry we got you covered. In this article, we will show you the top-down approach for getting started in applied machine learning.

Here’s What You Should Do Before You Get Started With Machine Learning

ML is all about applying statistics and computer science to data. You really do not need to be a professional programmer, mathematician to learn ML, but to master it, one has to be good at maths, programming and have some domain knowledge.

There are many programming languages which provide ML capabilities. But Python and R are most commonly used languages. So, before entering into the world of ML, choose one of these two programming languages – Python or R.

Python

Python is naturally disposed towards machine learning and is preferred by tech companies where they need end-to-end integration and develop analytics-based applications, leveraging analytics-friendly libraries. If you want more theoretical knowledge about different machine learning algorithms, you can also read Python Machine Learning Edition 2 written by a machine learning researchers Sebastian Raschka and Vahid Mirjalili. The book also covers large varieties of Practical Algorithms with Python, as well as using it with sci-kit-learn API and replaying it with Tensorflow API.

R

R as a language for statistical inference has made its name in data analysis and is preferred by companies which are primarily focused on advanced analytics and pretty much become a lingua franca for data science.

Learn Statistics For Machine Learning

It is good to have some understanding about statistics, especially the Bayesian probability, as it is essential for many machine learning algorithms. And to learn the basics of statistics, you can sign up for descriptive statistics and inferential statisticscourses offered by Udacity. Both the courses are free of cost.

ML Courses To Sharpen Your Knowledge

To build a strong machine learning foundation, soak in as much knowledge and theory as possible. There are various courses available to learn about machine learning:

Stanford’s Machine Learning Course:

It is a course for beginners that provides a broad introduction to machine learning, data mining and statistical pattern recognition. This course is taught by Andrew Ng and covers all basic algorithms. Topics include:

1. Supervised learning (parametric/non-parametric algorithms, support vector machines, kernels, neural networks)
2. Unsupervised learning (clustering, dimensionality reduction, recommender systems, deep learning)
3. Best practices in machine learning (bias/variance theory; innovation process in machine learning and AI)

The course will also draw from numerous case studies and applications so that you’ll also learn how to apply learning algorithms to building smart robots (perception, control), text understanding (web search, anti-spam), computer vision, medical informatics, audio, database mining, and other areas.

Neural Networks And Deep Learning Course

In this course, you will learn the foundation of deep learning and also teaches you how deep learning actually works. If you are looking for a job in AI, after this course you will also be able to answer basic interview questions. Once you enrol for a Certificate, you’ll have access to all videos, quizzes, and programming assignments (if applicable). This course is also taught by Andrew Ng.

Learning from Data Course By Professor Yaser Abu-Mostafa

This is an introductory course in ML that covers the basic theory, algorithms and applications. This ML course also balances theory and practice and covers the mathematical as well as heuristic aspects. However, this course is quite heavy on maths and requires more programming knowledge. The course is loaded with 8 homework sets.

Google’s Machine Learning Crash Course

Google’s Machine Learning crash course (MLCC) with TensorFlow APIs is a 15 hours online course that includes real-world case studies, interactive visualisation, video lectures, 40+ exercise to help teach machine learning concepts. Google originally designed this course for its employees as a part of a two-day boot camp aimed to give a practical introduction to machine learning fundamentals. More than 18,000 employees have already enrolled in MLCC, to enhance camera calibration for Daydream devices, build VR for Google Earth, and improve streaming quality at YouTube. Now, Google is making MLCC available to everyone.

Books To Go From Novice To Expert

Apart from the online courses, there are few good books available for machine learning. You can download the PDFs for your future use/reference:

It recommended for machine learning researchers and it gives a treatment of machine learning theory and mathematics.

This book presents approximate inference algorithms that permit fast approximate answers in situations where exact answers are not feasible. It uses graphical models to describe probability distributions.

It introduces a broad range of topics in deep learning. The text offers mathematical and conceptual background, covering relevant concepts in linear algebra, probability theory and information theory, numerical computation, and machine learning. It describes deep learning techniques used by practitioners in industry, including deep feedforward networks, regularisation, optimisation algorithms, convolutional networks, sequence modelling, and practical methodology; and it surveys such applications as natural language processing, speech recognition, computer vision, online recommendation systems, bioinformatics, and video games. This book is not available in PDF format but can order on Amazon.

Putting Theory Into Practice

Machine learning is more of an art, you can become good only by practising. For advanced level, you need to spend a lot of time working on various machine learning and deep learning problems. And you need datasets to practice building and for tuning models. You can start with UCI Machine Learning Repo or Kaggle.

UCI Machine Learning Repo

This contains 429 different datasets specially curated for practising machine learning. it has been widely used by students, educators, and researchers all over the world as a primary source of machine learning data sets. You can search by task, industry, dataset, size and more.

Kaggle

Kaggle is a great source of competitions and forums for ML hackathons and helps get one started on practical machine learning. It is an active, engaging and competitive platform and is more famous for hosting data science competition. Once you join Kaggle, don’t go on to expect to win competitions, but look at them as a way to gain real experience and mentoring from the community.

Build A Machine Learning Portfolio

Building a machine learning portfolio will go a long way in establishing how you can complete projects. It will also equip with the confidence to take on more interesting projects as you apply your ML learning and show your skills and capabilities to recruiters once you start looking for a job.

Here are a few project ideas:

1. A project where you had to collect the data yourself, e.g. scraping products reviews from a website
2. A project where you dealt with missing or messy data, e.g. cases where some people provided their location and some didn’t

Master In Specific Area That Will Help You Score A Job

Since machine learning is a broad field, it will be better to select a specific area of study and a Masters can help in landing a job interview as well. If you are geared towards a Post Doctorate which is also a good idea, there are a few application areas you can focus on. From Natural Language Processing to Computer Vision (think setting up GPU instances in AWS) and Deep Reinforcement Learning, there are several areas of application in ML for research.

And if you want to become more technically strong, deepen your Neural Network knowledge with a course or free resources that talks about artificial neural networks and how they’re being used for machine learning, in areas such as speech recognition, image segmentation and object recognition.

The post 5-Step Guide To Get Started In Machine Learning appeared first on Artificial Intelligence.

Google Analytics now has the same natural language processing technology available in other Google apps such as Photos and Search, the Internet giant announced on Tuesday. That means you can now ask questions in plain English, which gets you answers quicker than before.

Depending on the question you ask, you will be presented with a number, rows, or a chart. As an example, the Analytics team suggests a question such as “How many new users did we have from organic search on mobile last week?” which will give you a number. If you ask the trend of session duration, expect the answer in a chart form as the image above.

The feature becomes a part of Analytics Intelligence, which relies on machine learning to make sense of your analytics data. Analytics Intelligence will also help provide automated insights – now available on both Web and the app – alongside smart lists, smart goals, and session quality. Insights will also present specific recommendations to improve metrics, such as reducing load time to decrease bounce rate, and adding a new AdWords keyword to boost conversion rate.

To access the new questions feature and get automated insights, click the Intelligence button to open a side panel on the website, and tap the Intelligence icon in the top-right corner on the Google Analytics app for Android and iOS.

Google says the new features are now rolling out, and will be available in English to all Google Analytics users over the next few weeks.

The post Google Analytics Gets Natural Language Processing Support With Machine Learning appeared first on Artificial Intelligence.