Automotive, financial services, utilities--in these and many other industries, businesses are already applying artificial intelligence to core business processes and to innovating products. Business adoption of artificial intelligence is picking up steam, but still today only 20 percent of organizations that are aware of AI actually use this rapidly advancing technology. One reason: many executives are still wondering, "What can AI do for my business?" Earlier this year at the AI Frontiers conference in Santa Clara, California, we sat down with AI experts from some of the world's leading technology-first organizations to find out about current and future applications of AI. An edited version of the experts' remarks follows the video.

Companies across industries are experimenting with and using machine learning, but the actual adoption rates are lower than it might be seem. According to a 2017 SAP Digital Transformation Study, fewer than 10% of 3,100 executives from small, medium and large companies said their organizations were investing in machine learning. That will change dramatically in the coming years, according to a new Deloitte report, because researchers and vendors are making progress in five key areas that may make machine learning more practical for businesses of all sizes. There is a lot of debate about whether data scientists will or won't be automated out of a job. It turns out that machines are far better at doing rote tasks faster and more reliably than humans, such as data wrangling.

Artificial intelligence is not new but, suddenly, everyone seems to be talking about it. We have hit an inflection point with computing power and data that is finally allowing for commercial applications of this technology, and that's what all the excitement is about. It's only going to get faster and better from here on out. Along with talk about the new possibilities, there is also a lot of fear about people possibly losing their job to a robot, or even becoming irrelevant. Despite the wow factor of being able to shout a command at Siri or Alexa and have a task performed, when you get right down to it the tasks they are performing are rudimentary.

Artificial intelligence is not new but, suddenly, everyone seems to be talking about it. We have hit an inflection point with computing power and data that is finally allowing for commercial applications of this technology, and that's what all the excitement is about. It's only going to get faster and better from here on out. Along with talk about the new possibilities, there is also a lot of fear about people possibly losing their job to a robot, or even becoming irrelevant. Despite the wow factor of being able to shout a command at Siri or Alexa and have a task performed, when you get right down to it the tasks they are performing are rudimentary.

Deep learning (DL), a new-generation artificial neural network research, has made profound strides in recent years. This review paper is intended to provide water resources scientists with a simple technical overview, trans-disciplinary progress update, and potentially inspirations about DL. Effective architectures, more accessible data, advances in regularization, and new computing power enabled the success of DL. A trans-disciplinary review reveals that DL is rapidly transforming myriad scientific disciplines including high-energy physics, astronomy, chemistry, genomics and remote sensing, where systematic DL toolkits, innovative customizations, and sub-disciplines have emerged. However, with a few exceptions, its adoption in hydrology has so far been gradual. The literature suggests that novel regularization techniques can effectively prevent high-capacity deep networks from overfitting. As a result, in most scientific disciplines, DL models demonstrated superior predictive and generalization performance to conventional methods. Meanwhile, less noticed is that DL may also serve as a scientific exploratory tool. A new area termed "AI neuroscience", has been born. This budding sub-discipline is accumulating a significant body of work, e.g., distilling knowledge obtained in DL networks to interpretable models, attributing decisions to inputs via back-propagation of relevance, or visualization of activations. These methods are designed to interpret the decision process of deep networks and derive insights. While scientists so far have mostly been using customized, ad-hoc methods for interpretation, vast opportunities await for DL to propel advancement in water science.

Various applications involve assigning discrete label values to a collection of objects based on some pairwise noisy data. Due to the discrete---and hence nonconvex---structure of the problem, computing the optimal assignment (e.g.~maximum likelihood assignment) becomes intractable at first sight. This paper makes progress towards efficient computation by focusing on a concrete joint alignment problem---that is, the problem of recovering $n$ discrete variables $x_i \in \{1,\cdots, m\}$, $1\leq i\leq n$ given noisy observations of their modulo differences $\{x_i - x_j~\mathsf{mod}~m\}$. We propose a low-complexity and model-free procedure, which operates in a lifted space by representing distinct label values in orthogonal directions, and which attempts to optimize quadratic functions over hypercubes. Starting with a first guess computed via a spectral method, the algorithm successively refines the iterates via projected power iterations. We prove that for a broad class of statistical models, the proposed projected power method makes no error---and hence converges to the maximum likelihood estimate---in a suitable regime. Numerical experiments have been carried out on both synthetic and real data to demonstrate the practicality of our algorithm. We expect this algorithmic framework to be effective for a broad range of discrete assignment problems.

As IT leaders start mapping out their strategies for 2018, what trends should be top of mind and what challenges should organizations be prepared to face in the coming year? The value proposition for digital transformation has proven itself. According to Gartner, digital leaders perform better, generating more digital revenue and having more digital processes. Digital is becoming mainstream, and as a result the role of the CIO is changing. Top-performing digital organizations report that their CIOs have business responsibilities outside of IT, and this seat at the executive table is crucial for organizations leading digital transformation and adoption.

Product Owners, my field is failing you and everyone else. Gartner says A.I., Deep-Learning, and Machine Learning are at peak hype-cycle, and that is true when it comes to media reporting on A.I., but there is a way to cut through the chatter and get to value. And Product Owners, you are in the best position to move forward with A.I.. Imagine if instead of simply storing and retrieving images, audio, and video your software engineers could query that data and get back a resultset just like they do with your database. The above possibilities are the low-hanging fruit in an A.I. roll-out and adding the above to your product will pave the way for understanding and open up further possibilities for improving your product with A.I.. Let's take a step back and look at a very similar problem in the past; incorporating data into products. In the not so distant past, adding data to a product meant writing a giant check to Oracle, developing a custom schema, writing a connector in your language of choice, and writing raw SQL queries.

A project within the Stanford 100 Year Study on AI, The AI Index is an initiative to track, collate, distill and visualize data relating to artificial intelligence.

Labeling training data is increasingly the largest bottleneck in deploying machine learning systems. We present Snorkel, a first-of-its-kind system that enables users to train state-of-the-art models without hand labeling any training data. Instead, users write labeling functions that express arbitrary heuristics, which can have unknown accuracies and correlations. Snorkel denoises their outputs without access to ground truth by incorporating the first end-to-end implementation of our recently proposed machine learning paradigm, data programming. We present a flexible interface layer for writing labeling functions based on our experience over the past year collaborating with companies, agencies, and research labs. In a user study, subject matter experts build models 2.8x faster and increase predictive performance an average 45.5% versus seven hours of hand labeling. We study the modeling tradeoffs in this new setting and propose an optimizer for automating tradeoff decisions that gives up to 1.8x speedup per pipeline execution. In two collaborations, with the U.S. Department of Veterans Affairs and the U.S. Food and Drug Administration, and on four open-source text and image data sets representative of other deployments, Snorkel provides 132% average improvements to predictive performance over prior heuristic approaches and comes within an average 3.60% of the predictive performance of large hand-curated training sets.