A wide range of models have been proposed for Graph Generative Models, necessitating effective methods to evaluate their quality. So far, most techniques use either traditional metrics based onsubgraph counting, ortherepresentations of randomly initialized Graph Neural Networks (GNNs). We propose using representations from contrastively trained GNNs, rather than random GNNs, and show this gives more reliable evaluation metrics. Neither traditional approaches nor GNN-based approaches dominate the other,however: we giveexamples of graphs that each approach is unable to distinguish. We demonstrate that Graph Substructure Networks (GSNs), which in a way combine both approaches, are better at distinguishing the distances between graph datasets.

'Tis the season to slob out on the sofa and demolish a packet of mince pies in front of a good movie, or a bad movie - or a movie that's so bad it's good. This year, as ever, a crop of new Christmas films are hoping to be part of our festive viewing - and perhaps even join the ranks of enduring classics alongside the likes of Home Alone, Elf, Love Actually and Die Hard (don't start). So, in an effort to bring you a vital public service by sorting the crackers from the turkeys, and in an attempt to get myself into the Christmas spirit, I binged as many new Christmas films as possible in a day. This is the only 2025 release on Rotten Tomatoes' list of the greatest 100 Christmas movies of all time. The Jonas Brothers find themselves stuck in the UK after wrapping up their world tour and must get home for Christmas.

We introduce a method, MMD-B-Fair, to learn fair representations of data via kernel two-sample testing. We find neural features of our data where a maximum mean discrepancy (MMD) test cannot distinguish between representations of different sensitive groups, while preserving information about the target attributes. Minimizing the power of an MMD test is more difficult than maximizing it (as done in previous work), because the test threshold's complex behavior cannot be simply ignored. Our method exploits the simple asymptotics of block testing schemes to efficiently find fair representations without requiring complex adversarial optimization or generative modelling schemes widely used by existing work on fair representation learning. We evaluate our approach on various datasets, showing its ability to ``hide'' information about sensitive attributes, and its effectiveness in downstream transfer tasks.

Maximum mean discrepancy (MMD) is a particularly useful distance metric for differentially private data generation: when used with finite-dimensional features it allows us to summarize and privatize the data distribution once, which we can repeatedly use during generator training without further privacy loss. An important question in this framework is, then, what features are useful to distinguish between real and synthetic data distributions, and whether those enable us to generate quality synthetic data. This work considers the using the features of $\textit{neural tangent kernels (NTKs)}$, more precisely $\textit{empirical}$ NTKs (e-NTKs). We find that, perhaps surprisingly, the expressiveness of the untrained e-NTK features is comparable to that of the features taken from pre-trained perceptual features using public data. As a result, our method improves the privacy-accuracy trade-off compared to other state-of-the-art methods, without relying on any public data, as demonstrated on several tabular and image benchmark datasets.

Generations of computing professionals may remember Frederick P. Brooks, Jr., as the author of the seminal text on system engineering, The Mythical Man-Month: Essays on Software Engineeringa and his essays such as No Silver Bullet--Essence and Accident in Software Engineering.b Those who worked with Brooks, winner of the 1999 ACM A.M. Turing Award "for landmark contributions to computer architecture, operating systems, and software engineering," may also remember him as the lead designer of IBM's System/360, as an innovator in graphics and virtual reality, and as the founder of the University of North Carolina's computer science department. Brooks was born on April 19, 1931, in Greenville, North Carolina. He received his A.B. in Physics from Duke University in 1953. As a freshman, he saw an article in the January 23, 1950 issue of Time Magazine entitled "The Thinking Machine" that sparked his interest in computing.

Think AI is just coming for customer service jobs? Think again, say AI experts, who point to recent advances in large language models as evidence that white-collar and professional jobs will be disrupted too. Figuring out how AI and humans will coexist in the workplace is shaping up to be a key conversation for 2023 and beyond. "I think there are traditional white-collar businesses, white-collar professions that are going to be transformed by some of the innovation in large language models and AI technologies," said Peter Wang, the CEO of Anaconda, a provider of data science tools. "And that is going to create really interesting social and cultural dynamics that will basically settle out over the rest of this decade and reverberate into the 2030s."

Sutherland, an experience-led digital transformation company, announced that it has acquired AI-based customer experience platform company Augment CXM to extend Sutherland's AI solutions portfolio and help global brands improve efficiency, boost customer satisfaction and drive loyalty and conversion. Augment CXM's technology further supports Sutherland's commitment to deliver impactful customer journeys for global brands by enhancing the work of people with the very best in AI and intelligent automation. Sutherland's use of "Human-in-the-Loop" AI training models couples AI capabilities with human intelligence to optimize performance by providing real-time AI-based insights and suggested responses. "The addition of Augment CXM technology to our solutions expands and accelerates the innovative AI solutions we deliver," said Doug Gilbert, Chief Information Officer and Chief Digital Officer at Sutherland. "This powerful combination of augmenting human intelligence with AI technology to solve complex business problems enables our clients to achieve transformational business outcomes."

We are excited to bring Transform 2022 back in-person July 19 and virtually July 20 - 28. Join AI and data leaders for insightful talks and exciting networking opportunities. Over the past few years, we've all encountered "Let's chat!" buttons on websites that promise a quick, helpful customer service experience. But heavily hyped AI-driven chatbots, an important part of the customer experience mix since 2016, have also proven to be a mixed bag. Consumers found many bot interactions disappointing and time-consuming. Meanwhile, enterprises often needed to provide far more costly care and feeding of chatbots than expected.

More of the creative work these days is not being done by humans at all. When Dixons Carphone wanted to push shoppers towards its Black Friday sale, the company turned to Artificial Intelligence (AI) software and got the winning line "The time is now". Saul Lopes, head of customer marketing at Dixons Carphone, thinks it worked because it didn't have the words Black Friday in it. His human copywriters had produced dozens of potentially successful sentences but they all mentioned Black Friday. It was technology that broke this chain of thought.

Sentiment analysis is analytical technique that uses statistics, natural language processing, and machine learning to determine the emotional meaning of communications. Companies use sentiment analysis to evaluate customer messages, call center interactions, online reviews, social media posts, and other content. Sentiment analysis can track changes in attitudes towards companies, products, or services, or individual features of those products or services. Get the latest insights with our CIO Daily newsletter. One of the most prominent examples of sentiment analysis on the Web today is the Hedonometer, a project of the University of Vermont's Computational Story Lab.