Articles

Browse all AI articles. Use filters to narrow down results.

Showing 26881 to 26900 of 30062 entries.

Most Recent Articles

ArXiv Machine Learning
cs.lgcs.aics.clresearcharxiv machine learningllminference

Transparent Screening for LLM Inference and Training Impacts

By Arnault Pachot, Thierry Petit

arXiv:2604.19757v1 Announce Type: new Abstract: This paper presents a transparent screening framework for estimating inference and training impacts of current large language models under limited observability. The framework converts natural-language application descriptions into bounded environmental estimates and supports a comparative online observatory of current market models. Rather than claiming direct measurement for opaque proprietary services, it provides an auditable, source-linked proxy methodology designed to improve comparability, transparency, and reproducibility.

ArXiv Machine Learning
cs.lgcs.airesearcharxiv machine learningllmagentagents

arXiv:2604.19756v1 Announce Type: new Abstract: Large language model (LLM) agents often suffer from high reasoning overhead, excessive token consumption, unstable execution, and inability to reuse past experiences in complex tasks like business queries, tool use, and workflow orchestration. Traditional methods generate workflows from scratch for every query, leading to high cost, slow response, and poor robustness. We propose WorkflowGen, an adaptive, trajectory experience-driven framework for automatic workflow generation that reduces token usage and improves efficiency and success rate. Early in execution, WorkflowGen captures full trajectories and extracts reusable knowledge at both node and workflow levels, including error fingerprints, optimal tool mappings, parameter schemas, execution paths, and exception-avoidance strategies. It then employs a closed-loop mechanism that performs lightweight generation only on variable nodes via trajectory rewriting, experience updating, and template induction. A three-tier adaptive routing strategy dynamically selects among direct reuse, rewriting-based generation, and full initialization based on semantic similarity to historical queries. Without large annotated datasets, we qualitatively compare WorkflowGen against real-time planning, static single trajectory, and basic in-context learning baselines. Our method reduces token consumption by over 40 percent compared to real-time planning, improves success rate by 20 percent on medium-similarity queries through proactive error avoidance and adaptive fallback, and enhances deployability via modular, traceable experiences and cross-scenario adaptability. WorkflowGen achieves a practical balance of efficiency, robustness, and interpretability, addressing key limitations of existing approaches.

Techmeme
techtechmemegooglemeta

Connor Stringer / Telegraph: Trump threatens to “put a big tariff on the UK” if it does not drop its digital services tax, which he views as unfairly targeting US tech companies  —  President tells The Telegraph he will retaliate if Britain continues to ‘target’ major American companies such as Apple, Google and Meta

Techmeme
techtechmeme

Reuters: SAP reports Q1 revenue up 6% YoY to €9.56B, vs. €9.53B est., non-IFRS operating profit up 17% to €2.87B, cloud revenue up 19% to €5.96B; SAP up 10%+ after hours  —  Europe's largest software maker SAP (SAPG.DE) on Thursday reported a 17% increase in first-quarter profit …

Simon Willison Weblog
aisimon willison weblog

<p><strong><a href="https://github.com/russellromney/honker">russellromney/honker</a></strong></p> "Postgres NOTIFY/LISTEN semantics" for SQLite, implemented as a Rust SQLite extension and various language bindings to help make use of it.</p> <p>The design of this looks very solid. It lets you write Python code for queues that looks like this:</p> <pre><span class="pl-k">import</span> <span class="pl-s1">honker</span> <span class="pl-s1">db</span> <span class="pl-c1">=</span> <span class="pl-s1">honker</span>.<span class="pl-c1">open</span>(<span class="pl-s">"app.db"</span>) <span class="pl-s1">emails</span> <span class="pl-c1">=</span> <span class="pl-s1">db</span>.<span class="pl-c1">queue</span>(<span class="pl-s">"emails"</span>) <span class="pl-c1">emails</span>.<span class="pl-c1">enqueue</span>({<span class="pl-s">"to"</span>: <span class="pl-s">"alice@example.com"</span>}) <span class="pl-c"># Consume (in a worker process)</span> <span class="pl-k">async</span> <span class="pl-k">for</span> <span class="pl-s1">job</span> <span class="pl-c1">in</span> <span class="pl-s1">emails</span>.<span class="pl-c1">claim</span>(<span class="pl-s">"worker-1"</span>): <span class="pl-en">send</span>(<span class="pl-s1">job</span>.<span class="pl-c1">payload</span>) <span class="pl-s1">job</span>.<span class="pl-c1">ack</span>()</pre> <p>And Kafka-style durable streams like this:</p> <pre><span class="pl-s1">stream</span> <span class="pl-c1">=</span> <span class="pl-s1">db</span>.<span class="pl-c1">stream</span>(<span class="pl-s">"user-events"</span>) <span class="pl-k">with</span> <span class="pl-s1">db</span>.<span class="pl-c1">transaction</span>() <span class="pl-k">as</span> <span class="pl-s1">tx</span>: <span class="pl-s1">tx</span>.<span class="pl-c1">execute</span>(<span class="pl-s">"UPDATE users SET name=? WHERE id=?"</span>, [<span class="pl-s1">name</span>, <span class="pl-s1">uid</span>]) <span class="pl-s1">stream</span>.<span class="pl-c1">publish</span>({<span class="pl-s">"user_id"</span>: <span class="pl-s1">uid</span>, <span class="pl-s">"change"</span>: <span class="pl-s">"name"</span>}, <span class="pl-s1">tx</span><span class="pl-c1">=</span><span class="pl-s1">tx</span>) <span class="pl-k">async</span> <span class="pl-k">for</span> <span class="pl-s1">event</span> <span class="pl-c1">in</span> <span class="pl-s1">stream</span>.<span class="pl-c1">subscribe</span>(<span class="pl-s1">consumer</span><span class="pl-c1">=</span><span class="pl-s">"dashboard"</span>): <span class="pl-k">await</span> <span class="pl-en">push_to_browser</span>(<span class="pl-s1">event</span>)</pre> <p>It also adds 20+ custom SQL functions including these two:</p> <div class="highlight highlight-source-sql"><pre><span class="pl-k">SELECT</span> notify(<span class="pl-s"><span class="pl-pds">'</span>orders<span class="pl-pds">'</span></span>, <span class="pl-s"><span class="pl-pds">'</span>{"id":42}<span class="pl-pds">'</span></span>); <span class="pl-k">SELECT</span> honker_stream_read_since(<span class="pl-s"><span class="pl-pds">'</span>orders<span class="pl-pds">'</span></span>, <span class="pl-c1">0</span>, <span class="pl-c1">1000</span>);</pre></div> <p>The extension requires WAL mode, and workers can poll the <code>.db-wal</code> file with a stat call every 1ms to get as close to real-time as possible without the expense of running a full SQL query.</p> <p>honker implements the <strong>transactional outbox pattern</strong>, which ensures items are only queued if a transaction successfully commits. My favorite explanation of that pattern remains <a href="https://brandur.org/job-drain">Transactionally Staged Job Drains in Postgres</a> by Brandur Leach. It's great to see a new implementation of that pattern for SQLite. <p><small></small>Via <a href="https://news.ycombinator.com/item?id=47874647">Show HN</a></small></p> <p>Tags: <a href="https://simonwillison.net/tags/databases">databases</a>, <a href="https://simonwillison.net/tags/postgresql">postgresql</a>, <a href="https://simonwillison.net/tags/sqlite">sqlite</a>, <a href="https://simonwillison.net/tags/rust">rust</a></p>

Simon Willison Weblog
aisimon willison weblogclaudeanthropicllmagentagents

<p><strong><a href="https://www.anthropic.com/engineering/april-23-postmortem">An update on recent Claude Code quality reports</a></strong></p> It turns out the high volume of complaints that Claude Code was providing worse quality results over the past two months was grounded in real problems.</p> <p>The models themselves were not to blame, but three separate issues in the Claude Code harness caused complex but material problems which directly affected users.</p> <p>Anthropic's postmortem describes these in detail. This one in particular stood out to me:</p> <blockquote> <p>On March 26, we shipped a change to clear Claude's older thinking from sessions that had been idle for over an hour, to reduce latency when users resumed those sessions. A bug caused this to keep happening every turn for the rest of the session instead of just once, which made Claude seem forgetful and repetitive.</p> </blockquote> <p>I <em>frequently</em> have Claude Code sessions which I leave for an hour (or often a day or longer) before returning to them. Right now I have 11 of those (according to <code>ps aux  | grep 'claude '</code>) and that's after closing down dozens more the other day.</p> <p>I estimate I spend more time prompting in these "stale" sessions than sessions that I've recently started!</p> <p>If you're building agentic systems it's worth reading this article in detail - the kinds of bugs that affect harnesses are deeply complicated, even if you put aside the inherent non-deterministic nature of the models themselves. <p><small></small>Via <a href="https://news.ycombinator.com/item?id=47878905">Hacker News</a></small></p> <p>Tags: <a href="https://simonwillison.net/tags/ai">ai</a>, <a href="https://simonwillison.net/tags/prompt-engineering">prompt-engineering</a>, <a href="https://simonwillison.net/tags/generative-ai">generative-ai</a>, <a href="https://simonwillison.net/tags/llms">llms</a>, <a href="https://simonwillison.net/tags/anthropic">anthropic</a>, <a href="https://simonwillison.net/tags/coding-agents">coding-agents</a>, <a href="https://simonwillison.net/tags/claude-code">claude-code</a></p>

Business Insider
newsbusiness insiderchatgptclaudeanthropicmetaroboticsvision

Meta CEO Mark Zuckerberg. Arda Kucukkaya/Anadolu via Getty Images Tech job cut announcements have picked up again recently. Some factors pressuring tech jobs could be temporary, while others may be permanent. How long will this "AI air pocket" for tech jobs last? Multiple forces have converged to trigger a wave of job cuts across the tech industry. Some of these changes could be temporary, while others could stick and change the character of tech work forever. The latest news feels bad. Meta is cutting another 8,000 jobs. Microsoft offered buyouts to long-term employees. Oracle shed thousands of roles. Block chopped 40% of staff. These cuts are painful and stressful for the people involved. However, it pays to put recent numbers in context, gauge how much AI is to blame, and consider how permanent this situation might be. A recent report from Challenger, Gray & Christmas found that US tech layoffs were worse in 2023, only three years ago. Compared to the Great Recession of 2008 and 2009, this feels like a blip. I remember. I was there. It was terrifying. Here are some potentially temporary sources of pressure on tech jobs, along with others that could be permanent. Temporary pressure points Tech companies dramatically over-hired in 2020 and 2021 during the pandemic boom. They are still adjusting. Once that process is complete, job cuts could slow or even reverse. There's still demand for tech workers. Job postings for software development roles have risen sharply this year and have been climbing since late 2022, when ChatGPT launched. The supply of software engineers has risen dramatically, though, making the current situation feel worse, especially for some new graduates. Some tech giants are investing heavily in data centers and other AI infrastructure to develop new models and related services. That AI capex boom could peak around 2028, according to analyst projections. After that, some of these companies may have more financial flexibility, which could support a pickup in hiring. There is also what I'd call a potential "AI air pocket." Companies are experimenting with automation tools and pausing hiring as they evaluate their effectiveness. While AI has been blamed for job losses, it accounted for only about a quarter of cuts this year, according to Challenger estimates. As experiments conclude, some roles could return, especially where automation proves imperfect. Tech job losses are "not a result necessarily of AI successes," Gartner senior director analyst Kathy Ross said. "Instead, the layoffs seem to be part of a broader strategy to reinvest funds in AI, hoping for success down the line." Permanent job market changes Some types of jobs could disappear for good, particularly in areas like customer service, where AI systems can handle routine tasks more cheaply and efficiently. If you handled horses when the automobile arrived, then you probably lost your job. Should we have all kept riding horses instead of using cars and continued moving around slowly? Probably not. AI also increasingly rewards employees who build, create, and sell, rather than those focused on process and oversight. In Silicon Valley, that shift is showing up in flatter organizations, fewer middle managers, and a growing preference for employees who identify as "builders." AI tools can handle some routine work — analyzing internal data, coordinating workflows, or managing repetitive tasks — reducing the need for roles centered on supervision or process maintenance. By contrast, workers who produce tangible output can potentially use AI to get more things done more quickly. Comfortable roles that relied on maintaining and repeating established processes are being exposed as less essential. New kinds of tech jobs will emerge, and some of these may become long-term professions. AI tools like Claude Code are enabling a new "Design Producer" role in Silicon Valley, where senior individual contributors focus on creating, guiding, and elevating work, using judgment and taste to amplify team output, rather than managing people through traditional hierarchies. Then there are "robot wranglers," who move, fix, and train AI-powered robots. Sign up for BI's Tech Memo newsletter here. Reach out to me via email at abarr@businessinsider.com. Read the original article on Business Insider

Simon Willison Weblog
aisimon willison weblograg

<p><strong><a href="https://atproto.com/blog/serving-the-for-you-feed">Serving the For You feed</a></strong></p> One of Bluesky's most interesting features is that anyone can run their own <a href="bluesky custom feed">custom "feed" implementation</a> and make it available to other users - effectively enabling custom algorithms that can use any mechanism they like to recommend posts.</p> <p>spacecowboy runs the <a href="https://bsky.app/profile/did:plc:3guzzweuqraryl3rdkimjamk/feed/for-you">For You Feed</a>, used by around 72,000 people. This guest post on the AT Protocol blog explains how it works.</p> <p>The architecture is <em>fascinating</em>. The feed is served by a single Go process using SQLite on a "gaming" PC in spacecowboy's living room - 16 cores, 96GB of RAM and 4TB of attached NVMe storage.</p> <p>Recommendations are based on likes: what else are the people who like the same things as you liking on the platform?</p> <p>That Go server consumes the Bluesky firehose and stores the relevant details in SQLite, keeping the last 90 days of relevant data, which currently uses around 419GB of SQLite storage.</p> <p>Public internet traffic is handled by a $7/month VPS on OVH, which talks to the living room server via Tailscale.</p> <p>Total cost is now $30/month: $20 in electricity, $7 in VPS and $3 for the two domain names. spacecowboy estimates that the existing system could handle all ~1 million daily active Bluesky users if they were to switch to the cheapest algorithm they have found to work. <p>Tags: <a href="https://simonwillison.net/tags/go">go</a>, <a href="https://simonwillison.net/tags/scaling">scaling</a>, <a href="https://simonwillison.net/tags/sqlite">sqlite</a>, <a href="https://simonwillison.net/tags/software-architecture">software-architecture</a>, <a href="https://simonwillison.net/tags/tailscale">tailscale</a>, <a href="https://simonwillison.net/tags/bluesky">bluesky</a></p>

Techmeme
techtechmeme

Bloomberg: Sources: AI coding firm Cognition is in early talks to raise hundreds of millions of dollars at a $25B valuation, up from $10.2B announced in September 2025  —  Cognition AI Inc. is in early talks to raise a new round of funding that would more than double its valuation to $25 billion …

Business Insider
newsbusiness insiderragagentagents

Bloomberg/Getty Images US Army soldier Gannon Ken Van Dyke was involved in the planning of US action in Venezuela. Prosecutors say he used classified information to place trades on Polymarket and win over $400,000. Prosecutors announced his indictment on Thursday. A US Army soldier was indicted Thursday on charges he used classified military information to make bets on the prediction market platform Polymarket, winning more than $400,000. Gannon Ken Van Dyke, a Special Forces master sergeant who worked out of Fort Bragg in North Carolina, was charged with felonies including wire fraud and violating the Commodity Exchange Act, federal prosecutors in Manhattan announced. Van Dyke, prosecutors said in a statement, "was involved in the planning and execution of Operation Absolute Resolve," the military operation to capture Venezuela's Nicolas Maduro, "and had access to sensitive, nonpublic, classified information about that operation." Prosecutors say he used that information to place 13 bets between December 27, 2025, and January 2, 2026 — wagering more than $33,000 — in which he had an affirmative position on the question of whether the US would take action in Venezuela or against Maduro. Nicolas Maduro is escorted into a Manhattan courthouse after his capture. ADAM GRAY/Reuters He withdrew his winnings — a profit of $409,881 — the day of the US raid on Venezuela and deposited them first in a foreign cryptocurrency vault and then an online brokerage account, the indictment alleges. The feds say he also took steps to conceal his identity after speculation about the large, curiously timed trades began circulating, and asked Polymarket to delete his account using a phony excuse. Asked by a reporter if he was concerned about federal employees betting on prediction markets, President Donald Trump asked if Van Dyke had wagered that the US would get Maduro. When told that was the case, Trump said, "That's like Pete Rose betting on his own team," referring to the Cincinnati Reds manager who was banned from baseball for gambling. "If he bet against his team, that would be no good." No attorney for Van Dyke, 37, who has been an active-duty soldier since 2008, was listed in the court docket. A family member did not respond to a request for comment. In a post on X, Polymarket said it tipped off the feds to Van Dyke's trades. "Last month, we published our enhanced market integrity rules to combat insider trading. When we identified a user trading on classified government information, we referred the matter to the DOJ & cooperated with their investigation," the company wrote. "Insider trading has no place on Polymarket. Today's arrest is proof the system works." Insider trading is a major issue for prediction markets, which allow people to buy and sell contracts —in other words, trade — on the outcome of real-world events. On Wednesday, Polymarket rival Kalshi said it had suspended and fined three candidates for political office for making trades related to their own elections. In February, Kalshi banned and fined a MrBeast editor, who it said had traded on markets related to YouTube streaming. Van Dyke faces a far harsher penalty if convicted: The most serious charge against him, wire fraud, carries up to 20 years in prison. The Commodity Futures Trading Commission, which regulates prediction markets, filed a civil complaint against Van Dyke in federal court. "I have been crystal clear that anyone who engages in fraud, manipulation, or insider trading in any of our markets will face the full force of the law," Chairman Michael Selig said in a statement. "The defendant was entrusted with confidential information about US operations and yet took action that endangered US national security and put the lives of American service members in harm's way." Selig has defended prediction markets, saying in February, "It's something that I think is valuable to society." Read the original article on Business Insider

Wired Business
businessbusiness / regulationwired business

The master sergeant allegedly used classified intel to profit on the capture of Venezuelan president Nicolás Maduro, marking the first US arrest for insider trading on a prediction market.

Business Insider
newsbusiness insidervision

Nike to cut 1,400 jobs. Sheldon Cooper/SOPA Images/LightRocket via Getty Images Nike said it's cutting 1,400 roles in its global operations team, primarily in technology. Chief operating officer Venkatesh Alagirisamy announced the cuts in a memo on Thursday. The layoffs come as Nike's turnaround plan has taken longer than the company expected. Nike is benching some of its employees. The sportswear giant said Thursday that it is cutting about 1,400 roles in its global operations division, mainly in technology. Venkatesh Alagirisamy, who became chief operating officer in December, announced the reductions in a memo to staff detailing several changes "in the coming weeks." Alagirisamy wrote that the company is in the "final stretch" of the turnaround plan, dubbed its "win now" actions, which focuses on five key areas: culture, product, marketing, marketplace, and in-person presence. The changes include modernizing manufacturing, reallocating resources, and reshaping Nike's tech team, according to the memo. The cuts mark Nike's second round of cuts in 2026, following the elimination of 775 distribution center jobs. CEO Elliott Hill told analysts in a March earnings call that the turnaround plan is taking "longer than I would like." Read the full memo below: Team, NIKE, Inc. is in the final stretch of our Win Now action plan. Across the company, we have been taking deliberate steps to strengthen our foundation, sharpen how we compete, and build a model designed to deliver long-term profitable growth. Global Operations is critical to that effort. Today, I am sharing the next phase of that work and doing so with the entire company because of the scope and importance to our collective future. Over the coming months, we will continue evolving Global Operations to better serve athletes and the business with more speed, simplicity, and precision. Some of that work is happening now and more will continue over time as we align our teams, capabilities, and footprint to the future needs of the company. This is not a new direction. It is the next phase of the work already underway. Earlier this year, I outlined four actions we're taking across Operations to help make NIKE, Inc. a more responsive, resilient, responsible, and efficient company: optimizing our supply chain footprint, accelerating technology deployment, investing in upskilling our teams, and strengthening our partner and supplier relationships. The changes we are making now build on that work. At its core, these shifts are about building a Global Operations organization that is more focused, more integrated with the business, and better built for the pace of sport. In the coming weeks, several significant changes will take effect across Global Operations. These actions will result in changes to some team structures, work locations, and headcount. Technology Modernization. We're reshaping our Technology team to sharpen alignment with the business, build leaner teams, and accelerate what matters most. This means consolidating our technology footprint, streamlining our structure to move with greater speed and focus, and doubling down on the locations where our work gets done with a focus on two strategically important hubs — Philip H. Knight Campus and our Nike India Technology Center. These are deliberate moves to build a leaner, faster, more connected Technology organization — one built to deliver for the consumer, the athlete, and the business. Air Manufacturing Modernization. We are modernizing how work gets done across our Air Manufacturing Innovation (Air MI) facilities in Beaverton, Ore.; St. Louis; and Vietnam by increasing resiliency, streamlining processes, and adjusting staffing to match the needs of the business. The changes will help us stay flexible as demand changes, accelerate future innovations and continue to deliver the product quality athletes expect from Nike. Converse Ground Game. We are moving some Converse Footwear Manufacturing and Engineering resources closer to our factory partners to enable stronger real-time collaboration and to better support the brand's needs going forward. Materials Supply Chain Integration. We are bringing our Materials Supply Chain work more directly into our Footwear and Apparel Supply Chain teams while continuing to build on the valuable capabilities already in place. This strengthens how we operate from raw materials through finished product, enabling faster more synchronized decisions and ground-breaking innovations across our best-in-class manufacturing network. Collectively, these changes will result in a reduction of approximately 1,400 roles in Global Operations, with the majority in Technology. These reductions are very hard for the teammates directly affected and for the teams around them, too. Part of doing the right thing is being direct. Another part is making sure people are treated with care. Teammates whose roles are impacted will hear directly from their leaders and HR partners starting today, and we will work to make sure they have clear information and support through this transition. To teammates whose roles are impacted, thank you. Your contributions have helped build this company, and we are grateful for the work, commitment, and care you have brought to Nike. To the broader team, moments like this require professionalism, empathy, and a shared commitment to one another. These changes are meant to make the company less complex and more responsive. As we look ahead, that means simplifying parts of how we operate, using more advanced automation where it helps us work better, and building an even stronger end-to-end foundation for future growth. I expect these efforts to continue beyond our Win Now action plan. The Operations team has played, is playing, and will play an important role in helping NIKE, Inc. strengthen its foundation and it will continue to play a critical role in the company's comeback. Thank you for your leadership, your dedication to the athletes, and your dedication to each other. Venkatesh Alagirisamy EVP, Chief Operating Officer Read the original article on Business Insider

TechCrunch
startupslayoffsexclusiveredwood materialsjb straubeltechtechcrunch

Former Tesla executive Chris Lister is retiring, and at least three other VPs have left Redwood around the recent restructuring, TechCrunch has learned.