Restrepo (2025) develops a framework for economic growth in which Artificial General Intelligence (AGI) can perform any human task given sufficient computational resources. In his model, all economically essential "bottleneck" work is eventually automated, wages converge to the computational cost of replicating human work, and labor's share of GDP approaches zero as computational resources expand. This note relaxes one of his assumptions: that all task types have uniform automation costs. Drawing on Moravec's Paradox [1]--the observation that tasks humans find effortless (perception, mobility, manipulation) often require enormous computational resources, while tasks humans find difficult (mathematics, logic) require relatively modest computation--we extend his model to allow for differential automation costs across cognitive and physical tasks.

'The ludicrous valuations of AI startups are predicated on the idea that this technology has the power to eliminate the very need for human labor.' 'The ludicrous valuations of AI startups are predicated on the idea that this technology has the power to eliminate the very need for human labor.' AI will make the rich unfathomably richer. Is this really what we want? The'knowledge economy' promised cultural and social growth. Instead, we got worsening inequality and division. R ecently, Palantir - a tech corporation that boasts no fewer than five billionaire executives - announced its Q2 earnings: over a billion dollars generated in a single quarter.

Data contamination hinders fair LLM evaluation by introducing test data into newer models' training sets. Existing studies solve this challenge by updating benchmarks with newly collected data. However, they fail to guarantee contamination-free evaluation as the newly collected data may contain pre-existing knowledge, and their benchmark updates rely on intensive human labor. To address these issues, we in this paper propose AntiLeak-Bench, an automated anti-leakage benchmarking framework. Instead of simply using newly collected data, we construct samples with explicitly new knowledge absent from LLMs' training sets, which thus ensures strictly contamination-free evaluation. We further design a fully automated workflow to build and update our benchmark without human labor. This significantly reduces the cost of benchmark maintenance to accommodate emerging LLMs. Through extensive experiments, we highlight that data contamination likely exists before LLMs' cutoff time and demonstrate AntiLeak-Bench effectively overcomes this challenge.

The growing trend of artificial intelligence (AI) as a solution to social and technical problems reinforces AI Realism -- the belief that AI is an inevitable and natural order. In response, this paper argues that participatory design (PD), with its focus on democratic values and processes, can play a role in questioning and resisting AI Realism. I examine three concerning aspects of AI Realism: the facade of democratization that lacks true empowerment, demands for human adaptability in contrast to AI systems' inflexibility, and the obfuscation of essential human labor enabling the AI system. I propose resisting AI Realism by reconfiguring PD to continue engaging with value-centered visions, increasing its exploration of non-AI alternatives, and making the essential human labor underpinning AI systems visible. I position PD as a means to generate friction against AI Realism and open space for alternative futures centered on human needs and values.

Prompting methods have shown impressive performance in a variety of text mining tasks and applications, especially few-shot ones. Despite the promising prospects, the performance of prompting model largely depends on the design of prompt template and verbalizer. In this work, we propose MetricPrompt, which eases verbalizer design difficulty by reformulating few-shot text classification task into text pair relevance estimation task. MetricPrompt adopts prompting model as the relevance metric, further bridging the gap between Pre-trained Language Model's (PLM) pre-training objective and text classification task, making possible PLM's smooth adaption. Taking a training sample and a query one simultaneously, MetricPrompt captures cross-sample relevance information for accurate relevance estimation. We conduct experiments on three widely used text classification datasets across four few-shot settings. Results show that MetricPrompt outperforms manual verbalizer and other automatic verbalizer design methods across all few-shot settings, achieving new state-of-the-art (SOTA) performance.

IN2018, ON A TRIP to his ancestral homeland, Alexandr Wang listened as China's brightest engineers gave impressive presentations on artificial intelligence. He found it odd that the researchers conspicuously avoided any mention of how AI might be used. Wang, whose immigrant parents were nuclear physicists at Los Alamos National Laboratory, where the first atomic bombs were designed, was unsettled. "They were really dodgy on what the use cases were. You could tell it was for no good," recalls Wang, the cofounder of Scale AI, who has no second "e" in his first name so that it has eight characters, a number associated with good fortune in Chinese culture. Scale was then an up-and-coming startup providing data services primarily to self-driving auto-makers.

Reporting about AI is hard. When news articles uncritically repeat PR statements, overuse images of robots, attribute agency to AI tools, or downplay their limitations, they mislead and misinform readers about the potential and limitations of AI. We noticed that many articles tend to mislead in similar ways, so we analyzed over 50 articles about AI from major publications, from which we compiled 18 recurring pitfalls. We hope that being familiar with these will help you detect hype whenever you see it. We also hope this compilation of pitfalls will help journalists avoid them.

I have a grayish dual position regarding generative art and, well, basically, generative creativity. One view is extremely cynical, and the other perspective is hopeful. I wrote earlier about this topic here (note: a bit gloomy). Let me start with the cynical view, hyperbolized for ease of communication. I see this as a big tech effort to lower tech wages, reduce negotiation positions of creative workers, push the commoditization of art, create a new scaleable consumer market, and more holistically drive society towards transhumanism.

ChatGPT is all the rage. It is what everyone has been talking about in the last several weeks. In just over a week, it garnered over 1 million users, an incredible achievement for OpenAI, the organization that created it. ChatGPT is an Artificial Intelligence (AI) application that falls under the Generative AI category – GPT stands for Generative Pre-Trained Transformer. Generative AI enable computers to create new content using previously created content, such as text, audio, video, images and code.

Among the components of the World Economic Forum's Great Resetare a drastically reduced population and the replacement of human labor with robots and artificial intelligence (AI). The question immediately comes to mind: can robots and AI really make all the stuff for the elites after they have gotten rid of the people? Because a plan has been formulated and described does not mean that it is possible to realize. The plan may contradict laws of logic or reality, or assume the existence of resources that do not exist. Podcaster and journalist James Delingpole, speaking to investigative journalist Whitney Webb on October 23, 2021, discussed this topic with his guest. One of the main pillars of that is automation and artificial intelligence.