Frustration as lengthy Twitter outage hits Aussie users
GPT Takes the Bar Exam
Bommarito, Michael James and Katz, Daniel Martin, GPT Takes the Bar Exam (December 29, 2022). Available at SSRN: https://ssrn.com/abstract=4314839
“Nearly all jurisdictions in the United States require a professional license exam, commonly referred to as “the Bar Exam,” as a precondition for law practice. To even sit for the exam, most jurisdictions require that an applicant completes at least seven years of post-secondary education, including three years at an accredited law school. In addition, most test-takers also undergo weeks to months of further, exam-specific preparation. Despite this significant investment of time and capital, approximately one in five test-takers still score under the rate required to pass the exam on their first try. In the face of a complex task that requires such depth of knowledge, what, then, should we expect of the state of the art in “AI?” In this research, we document our experimental evaluation of the performance of OpenAI’s text-davinci-003 model, often-referred to as GPT-3.5, on the multistate multiple choice (MBE) section of the exam. While we find no benefit in fine-tuning over GPT-3.5’s zero-shot performance at the scale of our training data, we do find that hyperparameter optimization and prompt engineering positively impacted GPT-3.5’s zero-shot performance. For best prompt and parameters, GPT-3.5 achieves a headline correct rate of 50.3% on a complete NCBE MBE practice exam, significantly in excess of the 25% baseline guessing rate, and performs at a passing rate for both Evidence and Torts. GPT-3.5’s ranking of responses is also highly correlated with correctness; its top two and top three choices are correct 71% and 88% of the time, respectively, indicating very strong non-entailment performance. While our ability to interpret these results is limited by nascent scientific understanding of LLMs and the proprietary nature of GPT, we believe that these results strongly suggest that an LLM will pass the MBE component of the Bar Exam in the near future.”
Teri Kanefield’s long read is an articulate, in-depth analysis of the rapid, jarring, deeply troubling shifts occurring on social media giant Twitter, and the choices she recommends users should make moving forward.
The New York Times: “Twitter is said to have stopped paying rent at its Seattle office, leading it to face eviction. Janitorial and security services have been cut, and in some cases employees have resorted to bringing their own toilet paper to the office...
Elon Musk has reduced the company to a bare-bones operation, and employees are under a “zero-based budgeting” mandate to justify any spending. Early on Christmas Eve, members of the billionaire’s staff flew to Sacramento — the site of one Twitter’s three main computing storage facilities — to disconnect servers that had kept the social network running smoothly. Some employees were worried that losing those servers could cause problems, but saving money was the priority, according to two people who were familiar with the move but not authorized to talk about it.
The data center shutdown was one of many drastic steps Mr. Musk has undertaken to stabilize Twitter’s finances. Over the past few weeks, Twitter had stopped paying millions of dollars in rent and services, and Mr. Musk had told his subordinates to renegotiate those agreements or simply end them. The company has stopped paying rent at its Seattle office, leading it to face eviction, two people familiar with the matter said. Janitorial and security services have been cut, and in some cases employees have resorted to bringing their own toilet paper to the office. Mr. Musk bought the social network for $44 billion in late October, saddling it with debt that will require him to pay about $1 billion in interest annually. Speaking on a live forum on Twitter last week, Mr. Musk compared the company to a “plane that is headed towards the ground at high speed with the engines on fire and the controls don’t work.” Twitter was on track to have a “negative cash flow situation” of about $3 billion in 2023, he said, citing a depressed advertising environment and increased costs, like the debt payments…”
Mastodon and the pros and cons of moving beyond Big Tech gatekeepers Ars Technica: “As Elon Musk’s Category 5 tweetstorm continues, the once-obscure Mastodon social network has been gaining over 1,000 new refugees per hour, every hour, bringing its user count to about eight million. Joining as a user is pretty easy. More than enough ex-Twitterers are happy finding a Mastodon instance via joinmastodon.org, getting a list of handles for their Twitter friends via Movetodon, and carrying on as before. But what new converts may not realize is that Mastodon is just the most prominent node in a much broader movement to change the nature of the web. With a core goal of decentralization, Mastodon and its kin are “federated,” meaning you are welcome to put up a server as a home base for friends and colleagues (an “instance”), and users on all instances can communicate with users on yours. The most common metaphor is email, where yahoo.com, uchicago.edu, and condenast.com all host a local collection of users, but anybody can send messages to anybody else via standard messaging protocols. With cosmic ambitions, the new federation of freely communicating instances is called “the Fediverse.” I started using Mastodon in mid-2017 when I faintly heard the initial buzz. I found that the people who inhabited a world whose first major selling point was its decentralized network topology were geeky and countercultural. There were no #brands. Servers were (and are) operated by academic institutions, journalists, hobbyists, and activists in the LGBTQ+ community. The organizers of one instance, scholar.social, run an annual seminar series, where I have presented. The decentralization aspect that was such a selling point for me was also a core design goal for Mastodon and the predecessors it built upon, such as GNU Social. In an interview with Time, lead developer Eugen Rochko said that he began development of Mastodon in 2016 because Twitter was becoming too centralized and too important to discourse. “Maybe it should not be in the hands of a single corporation,” he said. His desire to build a new system “was generally related to a feeling of distrust of the top-down control that Twitter exercised.” [Note: connect with beSpacific on Mastodon – updated daily!]
…As with many a web app, Mastodon is a duct taping together of components and standards; hosting or interacting with a Mastodon instance requires some familiarity with all of these. Among them, and the headliner at the heart of The Fediverse, is the ActivityPub standard of the World Wide Web Consortium (W3C), which specifies how actors on the network are defined and interact…”
