Salesforce has slashed another 4,000 jobs from its customer support workforce as the tech giant doubles down on artificial intelligence, even as the company reports strong financial results. AI agents now reportedly handle about one million customer conversations. In a recent episode of The Logan Bartlett Show, CEO Marc Benioff justified the cuts by saying he "needs less heads" as Salesforce invests heavily in AI across its operations. Earlier this year, Benioff boasted that AI was already doing 30 to 50 percent of the work, which he framed as efficiency gains – a 17 percent cost reduction achieved after shedding 1,000 people in February. On Wednesday, the Slack owner reported revenue topped 10.2bn for the quarter ending July 31, up 10 percent from the same period last year.

By reinterpreting a robust discriminative classifier as Energy-based Model (EBM), we offer a new take on the dynamics of adversarial training (AT). Our analysis of the energy landscape during AT reveals that untargeted attacks generate adversarial images much more in-distribution (lower energy) than the original data from the point of view of the model. Conversely, we observe the opposite for targeted attacks. On the ground of our thorough analysis, we present new theoretical and practical results that show how interpreting AT energy dynamics unlocks a better understanding: (1) AT dynamic is governed by three phases and robust overfitting occurs in the third phase with a drastic divergence between natural and adversarial energies (2) by rewriting the loss of TRadeoff-inspired Adversarial DEfense via Surrogate-loss minimization (TRADES) in terms of energies, we show that TRADES implicitly alleviates overfitting by means of aligning the natural energy with the adversarial one (3) we empirically show that all recent state-of-the-art robust classifiers are smoothing the energy landscape and we reconcile a variety of studies about understanding AT and weighting the loss function under the umbrella of EBMs. Motivated by rigorous evidence, we propose Weighted Energy Adversarial Training (WEAT), a novel sample weighting scheme that yields robust accuracy matching the state-of-the-art on multiple benchmarks such as CIFAR-10 and SVHN and going beyond in CIFAR-100 and Tiny-ImageNet. We further show that robust classifiers vary in the intensity and quality of their generative capabilities, and offer a simple method to push this capability, reaching a remarkable Inception Score (IS) and FID using a robust classifier without training for generative modeling. The code to reproduce our results is available at http://github.com/OmnAI-Lab/Robust-Classifiers-under-the-lens-of-EBM/ .

When Shahmeer Mirza began working at PepsiCo., which owns the Frito-Lay brand, one of his first tasks was teaching a machine sensory perception. By bouncing lasers off chips, he could capture the sound they make when someone bites into one. The thickness of the chip and the time and temperature of the frying could all be adjusted to create the perfect bite without a worker having to constantly sample chips. It was such a revolution his work was awarded a patent, but it's only one way in which A.I. is being put to use at snack plants. Repetitive tasks like picking out a bad potato and throwing it away might take a worker a second or two.

One business who realized that using artificial intelligence (AI) and machine learning is a business need, no longer a competitive advantage is PepsiCo. The food-and-beverage company behind brands such as Pepsi, Gatorade, Tropicana, Lipton, Frito-Lay, and Quaker sells products in more than 200 countries and brought in $64.7 billion in annual revenue last year. From robots to machine learning, PepsiCo uses AI and machine learning throughout the organization in many ways. There's a six-wheeled mobile vending machine robot tooling around the University of the Pacific chockful of PepsiCo snacks and beverages from Hello Goodness--a healthier line-up that includes SunChips, Baked Lay's and bubly sparkling water. Named Snackbot, these self-driving robots are a partnership between Robby Technologies and PepsiCo.

A "sexual predator" has been jailed for eight and a half years for blackmailing and spying on Muslim women using intimate photographs and videos he took of them without their knowledge. Farhan Mirza, 38, of Abertillery, Blaenau Gwent, secretly filmed the women and threatened to share the footage before demanding money. Mirza, who denied the charges, met some of the women on online dating sites. He was jailed for voyeurism, blackmail, theft and fraud at Cardiff Crown Court. During the trial, jurors heard Mirza had initially impressed his victims by claiming to be a doctor and hung surgical scrubs in his wardrobe and carried a stethoscope in his car. He also claimed his family were highly educated professionals working in locations around the world.

This paper introduces the concept and use of an Interaction History Architecture for use on a mobile domestic robot and analyses the limitations of this configuration. The interaction history architecture builds upon Shannon information theory and has been previously used in a humanoid robot to learn basic children’s games. Previous work has shown that experience spaces can be highly flexible when used for learning. In this paper we outline and experiment designed to test the abilities of the architecture and how it can be used with classic clicker style training to teach domestic robots simple tasks. It then presents results from an experiment exploring these capabilities as well as the limitation found therein.