Training a state-of-the-art Large Language Model (LLM) is an increasingly expensive endeavor due to growing computational, hardware, energy, and engineering demands. Yet, an often-overlooked (and seldom paid) expense is the human labor behind these models' training data. Every LLM is built on an unfathomable amount of human effort: trillions of carefully written words sourced from books, academic papers, codebases, social media, and more. This position paper aims to assign a monetary value to this labor and argues that the most expensive part of producing an LLM should be the compensation provided to training data producers for their work. To support this position, we study 64 LLMs released between 2016 and 2024, estimating what it would cost to pay people to produce their training datasets from scratch. Even under highly conservative estimates of wage rates, the costs of these models' training datasets are 10-1000 times larger than the costs to train the models themselves, representing a significant financial liability for LLM providers. In the face of the massive gap between the value of training data and the lack of compensation for its creation, we highlight and discuss research directions that could enable fairer practices in the future.

Artificial intelligence executives have big plans--and they're not cheap. In a recent interview with TIME, Dario Amodei, CEO of AI company Anthropic predicted that the cost to develop the next generation of AI systems that will be released later this year would be around 1 billion. This trend suggests that the generation after that would cost more like 10 billion. Amodei is not the only one preparing for a spending spree. Microsoft and OpenAI are reportedly planning to build a 100 billion supercomputer to build and run AI models.

It didn't take long for Harshraj Ghai to respond to the impact of California's new 20 an hour minimum wage for his 3,700 fast-food employees. Ghai and his family operate 180 Burger Kings, Taco Bells and Popeyes chicken restaurants across the state, and one of the first things they did after the law took effect April 1 was to start capping workers' hours to avoid overtime pay. Also, they're closing some outlets a little earlier, and opening others a bit later to avoid paying workers for less profitable periods. But the biggest thing Ghai and his family are doing does not directly involve workers at all: They've speeded up and expanded their use of technology, especially AI. With the state's mandatory minimum wage for fast-food workers set to increase to 20 an hour, many restaurant chains are preparing to raise prices.

Teleoperated or remote-controlled driving complements automated driving and acts as transitional technology toward full automation. An economic advantage of teleoperated driving in logistics operations lies in managing fleets with fewer teleoperators compared to vehicles with in-vehicle drivers. This alleviates growing truck driver shortage problems in the logistics industry and saves costs. However, a trade-off exists between the teleoperator-to-vehicle ratio and the service level of teleoperation. This study designs a simulation framework to explore this trade-off generating multiple performance indicators as proxies for teleoperation service level. By applying the framework, we identify factors influencing the trade-off and optimal teleoperator-to-vehicle ratios under different scenarios. Our case study on road freight tours in The Netherlands reveals that for any operational setting, a teleoperation-to-vehicle ratio below one can manage all freight truck tours without delay, while one represents the current situation. The minimum teleoperator-to-vehicle ratio for zero-delay operations is never above 0.6, implying a minimum of 40% teleoperation labor cost saving. For operations where a small delay is allowed, teleoperator-to-vehicle ratios as low as 0.4 are shown to be feasible, which indicates potential savings of up to 60%. This confirms great promise for a positive business case for the teleoperated driving as a service.

There's no denying that the robotics startup world has taken a hit during the ongoing economic downturn. Recent numbers prove what we've all suspected for some time. But two things are true: 1) The lull is temporary; and 2) While robotics isn't recession-proof, construction might as well be. This is certainly a theme of late -- as other categories of robotics have struggled to raise, those operating in construction appear relatively unimpacted. New York-based Toggle this morning announced that it has added another $3 million to its coffers as part of a "Series A Extension."

However, because food inflation was both sharp and sustained, these businesses found their customers willing to accept price increases–and, because inflation was widespread across the industry, those who acted swiftly experienced better performance. In fact, businesses that raised prices sooner rather than later found themselves in a more advantageous position than those who held back. Now that commodity prices have fallen and runaway inflation has started to subside, food, beverage, and agribusiness companies–even those that didn't raise prices–may have some breathing room on the horizon. The food industry is starting to see light at the end of the tunnel despite continued high labor costs and lingering pandemic-related operational disruptions. Food industry margins have been reinforced by a number of post-pandemic shifts that are adding flexibility to the value chain.

Automation covers technologies across many processes and fields. In regard to customer service it's using technologies instead of people to accomplish both customer facing and back end tasks. Customers are leaning into automation according to McKinsey research. Though the majority surveyed (79%) use the telephone to connect to service, it's also the channel that most (75%) don't want to use in the future. Instead, 81% are more interested in using email and 62% in website/self-service in the future.

But President Xi Jinping is engaged in a zero-sum game, similar to Soviet leaders. "For long, Xi's zero-sum instincts and his entourage, no different than that of Communist leaders from the Soviet era, were successful in navigating the Western liberal rules-based system of trade and international institutions," he told International Business Times in an email. "Recent events (e.g., the opaqueness regarding the pandemic's origins, the war in Ukraine, and China's non-commitment to greenhouse commitments) and repeated threats to Taiwan appear to have spooked even the most dovish U.S. administration in recent memory." Moreover, Professor Colares thinks Xi Jinping's Soviet-style leadership has not helped China overcome the middle-income trap. In this situation, an emerging market economy slows down, failing to transition to a high-income and become a developed economy.

Aiming to ensure chatbot quality by predicting chatbot failure and enabling human-agent collaboration, Machine-Human Chatting Handoff (MHCH) has attracted lots of attention from both industry and academia in recent years. However, most existing methods mainly focus on the dialogue context or assist with global satisfaction prediction based on multi-task learning, which ignore the grounded relationships among the causal variables, like the user state and labor cost. These variables are significantly associated with handoff decisions, resulting in prediction bias and cost increasement. Therefore, we propose Causal-Enhance Module (CEM) by establishing the causal graph of MHCH based on these two variables, which is a simple yet effective module and can be easy to plug into the existing MHCH methods. For the impact of users, we use the user state to correct the prediction bias according to the causal relationship of multi-task. For the labor cost, we train an auxiliary cost simulator to calculate unbiased labor cost through counterfactual learning so that a model becomes cost-aware. Extensive experiments conducted on four real-world benchmarks demonstrate the effectiveness of CEM in generally improving the performance of existing MHCH methods without any elaborated model crafting.

The longstanding practice of watching hundreds and thousands of cameras for suspicious behavior, and then reacting, is over. This method has proven ineffective, especially as surveillance continues to proliferate in home, business, smart cities and other connected environments. In addition, conventional live-video monitoring services tend to not mention the total inherent delay times from the detection of an intrusion to the execution of effective deterrence reactions. With the ongoing shortage of labor and contract guard services, as well as humans who simply can't stay attentive to multiple video displays, technology is stepping up to assist and revolutionize remote monitoring services. Sensing, detection, analytics and artificial intelligence (AI) have changed the formula of monitoring from an after-the-fact forensic activity to a proactive and strategic tool that can actually deter and prevent crime and property loss, at a lower total cost of ownership to the user.