Nicole Hennig (@nic221) — Bluesky Profile

15 hours ago

Text Shot: "Seeing the agent do this entire workflow end-to-end and all by itself... is wild," Karpathy remarked, noting that the agent caught oversights in attention scaling and regularization that he had missed manually over two decades of work.

This is more than just a productivity hack; it is a fundamental shift in how intelligence is refined. By automating the "scientific method" for code, Karpathy has turned machine learning into an evolutionary process that runs at the speed of silicon rather than the speed of human thought.

And more than this, it showed the broader AI and machine learning community on X that this type of process could be applied far beyond computer science, to fields like marketing, health, and, well, basically anything that requires research.

Andrej-karpathys-new-open-source-autoresearch-lets-you-run-hundreds-of-ai venturebeat.com/technology/and… #AI #agents

2 0 1 0

15 hours ago

Text Shot: SEARCH AI was developed in response to changing research behavior. Library staff noticed that patrons were increasingly entering full questions or conversational phrases when searching the catalog, a pattern influenced by the growing presence of large language models and conversational search tools in everyday digital platforms.

Traditional library systems work best if users have familiarity with keyword searching, Boolean logic and controlled vocabulary. For many students, that can make it more difficult to find relevant sources.

University Libraries Named Finalist for Innovation Award news.stonybrook.edu/university/sto… #AI #libraries #search

2 0 2 0

15 hours ago

Text Shot: The owner of the data center is committed to building with sustainability in mind. By adding clean and renewable energy to the project, this new data center will support environmental goals and bring power to the community.

Water stewardship is another key priority. The data center is designed to use less water than the site does today, with a goal of restoring more water than it consumes. The owner is partnering on local restoration projects to improve water quality and investing in community organizations to address shared water challenges in the Mississippi River basin.

Richland Parish Data Center www.mortenson.com/projects/richl… #AI #DataCenters

2 0 0 0

1 day ago

Text Shot: Microsoft on Monday announced it is integrating Claude into Microsoft 365 Copilot through a new product called Copilot Cowork, developed in close collaboration with Anthropic. As Yahoo Finance reported, the service enables enterprise users to perform tasks like building presentations, pulling data into Excel spreadsheets, and coordinating meetings — the kind of agentic productivity capabilities that sent shares of SaaS companies like Salesforce, ServiceNow, and Intuit tumbling when Anthropic first debuted its Cowork product on January 30.

The timing is not coincidental. As TechCrunch reported last week, Microsoft, Google, and Amazon Web Services all confirmed that Claude remains available to their customers for non-defense workloads. Microsoft's legal team specifically concluded that "Anthropic products, including Claude, can remain available to our customers — other than the Department of War — through platforms such as M365, GitHub, and Microsoft's AI Foundry."

That…

Anthropic-rolls-out-code-review-for-claude-code-as-it-sues-over-pentagon venturebeat.com/technology/ant… #AI #Anthropic #Microsoft

1 0 0 0

1 day ago

Text Shot: Meta has acquired Moltbook, a viral social network designed for AI agents, Axios has learned.

Why it matters: The deal brings Moltbook's creators — Matt Schlicht and Ben Parr — into Meta Superintelligence Labs (MSL), the unit run by former Scale AI CEO Alexandr Wang.
Meta did not disclose Moltbook's purchase price.
The deal is expected to close mid-March, Meta says, with the pair starting at MSL on March 16.

Meta acquires Moltbook, the social network for AI agents www.axios.com/2026/03/10/met… #AI #Moltbook

1 0 0 0

1 day ago

Text Shot: NVIDIA IS PLANNING to launch an open-source platform for AI agents, people familiar with the company’s plans tell WIRED.

The chipmaker has been pitching the product, referred to as NemoClaw, to enterprise software companies. The platform will allow these companies to dispatch AI agents to perform tasks for their own workforces. Companies will be able to access the platform regardless of whether their products run on Nvidia’s chips, sources say.

Nvidia Is Planning to Launch an Open-Source AI Agent Platform www.wired.com/story/nvidia-p… #AI #agents #OpenSource

4 2 1 1

2 days ago

What LLMs allow is for one to set a set of objectives and guiding principles and desired outputs and let them handle the instances I haven't thought about without getting into the rigidness of code. I can say look, the point is for this to be accessible to screen-readers, and if it hits a landscape mode page it knows it needs to recreate that HTML as a "flow" and not as precise positioning. If it hits handwriting, it can just use its vision to read the handwriting.
If I tell it to write something back into a PDF it can use PyMuPdf, but if it hits a particular problem that PyMuPdf can't solve it can look for something else.

Softerware - Mike Caulfield mikecaulfield.substack.com/p/softerware #AI #skills #Claude

2 0 1 0

3 days ago

Text Shot: Since launching Agent Skills last October, we've noticed that most authors are subject matter experts, not engineers. They know their workflows but don't have the tools to tell whether a skill still works with a new model, triggers when it should, or if it actually improved after an edit.

Today we're announcing skill-creator enhancements that help authors build with more confidence. We are bringing some of the rigor of software development (testing, benchmarking, iterative improvement) to skill authoring without requiring anyone to write code.

Improving skill-creator: Test, measure, and refine Agent Skills | Claude claude.com/blog/improving… #AI #Claude #skills

5 1 1 1

3 days ago

Text Shot: The tool, whose documentation describes it as “one CLI for all of Google Workspace, built for humans and AI agents,” is called gws. It provides unified command-line access to Gmail, Google Drive, Calendar, Docs, Sheets, Slides, Chat, and most other Workspace services.

But the more revealing detail is buried in the instructions: the documentation includes a dedicated integration guide for OpenClaw, the open-source AI agent that went viral in late January and has since become something of a Rorschach test for where agentic AI is headed.

Google made Gmail and Drive easier for AI agents to use thenextweb.com/news/google-ma… #AI #Google #agents

3 1 1 0

3 days ago

Text Shot: To make your own tool or dashboard, select the new Canvas option from the tool menu (+) in AI Mode and describe what you want to create. You’ll get a working prototype in the Canvas side panel that pulls together the freshest information from the web and Google’s Knowledge Graph. From there, you can test the functionality, toggle to view the underlying code and refine with conversational follow-ups until it does exactly what you need.

Use Canvas in AI Mode to get things done and bring your ideas to life, right in Search. - Google blog.google/products-and-p… #AI #Google #VibeCoding

2 0 0 0

3 days ago

Text Shot: Run LTX Desktop on your own compute with full access to model weights. No internet required. No cost per generation. Install, modify, and deploy on your own infrastructure.
Free for individuals and companies under $10M in annual revenue.

Free AI Video Production Suite: Generate & Edit Videos Online | LTX Desktop ltx.io/ltx-desktop #AI #video #OpenSource

3 0 0 0

3 days ago

Text Shot: The BBC World Service programme Witness History will release five animated video adaptations of recent episodes, created using generative AI technology. This marks a first for the BBC World Service, as it explores new ways to bring its journalism to audiences worldwide.

The first animated episode, The World’s First Labradoodle, is now available on the BBC World Service YouTube channel. It will be followed by animated versions of Brazil’s Biggest Bank Heist; Ramesses II’s ‘Mummy Makeover’; The Discovery of Lord Sipan in Peru; and Arrested for Playing Football in Brazil.

The project aims to extend the reach of Witness History by transforming its acclaimed audio storytelling into an engaging visual experience.

BBC World Service’s Witness History to launch first AI-animated video episodes www.bbc.co.uk/mediacentre/20… #AI #video #journalism

1 0 0 0

4 days ago

Text Shot: Americans are more likely than people in other countries surveyed in 2025 to question the morality of their fellow countrymen, according to Pew Research Center surveys in 25 countries.

In 25-Country Survey, Americans Especially Likely To View Fellow Citizens as Morally Bad | Pew Research Center www.pewresearch.org/religion/relig… #surveys

0 0 0 0

4 days ago

The most interesting thing happening in Al right now isn't the models, it's the harnesses acting as the integration layer: tool calling, UX, agentic orchestration. A great model in a mediocre harness loses to a good model in a great harness. Gemini's models are competitive but feel underwhelming because Google's tooling can't showcase them, which is presumably why they acquired Windsurf and relaunched it as Antigravity. Claude's models shine brightest through Claude Code. The engine matters, but nobody buys an engine.
Hot take: The model is the engine. The harness is the car.

The Harness Is The Product (and Other Hot Takes) : Dave Beckett www.dajobe.org/blog/2026/03/0… man #coding #agents

1 0 0 0

5 days ago

Vibe Coding for Beginners. A Library 2.0 masterclass with Nicole Hennig

VIBE CODING FOR BEGINNERS www.library20.com/ai-sessions/vi… (here’s my upcoming workshop for librarians, join us on Mach 20th) #AI #webinars #librarians

2 1 0 0

5 days ago

Text Shot: The big headlines on this release are efficiency, with OpenAI reporting that GPT-5.4 uses far fewer tokens (47% fewer on some tasks) than its predecessors, and, arguably even more impressively, a new "native" Computer Use mode available through the API and its Codex that lets GPT-5.4 navigate a users' computer like a human and work across applications.

The company is also releasing a new suite of ChatGPT integrations allowing GPT-5.4 to be plugged directly into users' Microsoft Excel and Google Sheets spreadsheets and cells, enabling granular analysis and automated task completion that should speed up work across the enterprise, but may make fears of white collar layoffs even more pronounced on the heels of similar offerings from Anthropic's Claude and its new Cowork application.

Openai-launches-gpt-5-4-with-native-computer-use-mode-financial-plugins-for venturebeat.com/technology/ope… #AI #OpenAI

2 0 0 0

5 days ago

Text Shot: In this work, we study whether current reasoning models are capable of controlling their chain of thought in ways that reduce monitorability. Understanding this capability is important for ensuring that CoT monitoring remains a robust safeguard as AI systems grow more capable. We find that current reasoning models struggle to control their CoTs, even when told they are being monitored. While controllability is higher for larger models, it decreases as models are asked to reason for longer and when they undergo additional post-training. Our results suggest that CoT controllability is unlikely to currently pose a major risk to CoT monitorability, though continued evaluation will be important as models advance.

Reasoning models struggle to control their chains of thought, and that’s good openai.com/index/reasonin… #AI #OpenAI

0 0 0 0

5 days ago

Text Shot: As the saying goes, “a lie can travel halfway around the world while the truth is putting on its shoes”. There is a small kernel of truth here, but this is also hugely misleading, proving that misinformation often spreads faster than the truth.

The reality is that the US Supreme Court denied certiorari in the case of Thaler v Perlmutter (more on the details of that case later). This doesn’t mean that the SCOTUS ruled on the issue, in fact the decision is quite simply a line in the corresponding docket, which makes most of the reporting inaccurate. Because the Supreme Court declined to intervene, the decision made by the last court to hear the case, in this case the U.S. Court of Appeals for the District Court of Columbia Circuit, becomes the final word for the parties involved. There are generally no further avenues for appeal.

Also important for most of the erroneous reporting out there, this decision does not set any precedent, which is a very common misconception.

No, the US Supreme Court did not declare that AI works cannot be copyrighted www.technollama.co.uk/no-the-us-supr… #AI #cooyright

3 0 0 0

5 days ago

Text Shot: Our knowledge of the past comes from stone tablets and old parchment. But thousands of years from now, our descendants may learn of our lives from a thin slice of glass carrying an impressive load of data—all thanks to physics that sounds borderline magical.

Today, Microsoft’s Project Silica unveiled the latest technological advances in what is essentially laser-modified glass storage for sensitive data. Described in a Nature paper, the sy

Microsoft's Glass Chip Holds Terabytes of Data for 10,000 Years gizmodo.com/microsofts-gla… #preservation #storage #future

2 1 0 0

5 days ago

Text Shot: In 2026, all net new generating capacity = renewables
Capacity growth from utility-scale renewables and batteries in 2026 is projected at 80,809.2 MW. (EIA does not provide a forecast for small-scale solar, but, based on recent growth rates, the SUN DAY Campaign estimates it will provide an additional 6,000 MW or more.)
The net capacity of natural gas, coal, and oil is forecast to fall by 4,211.6 MW. No new nuclear capacity is currently predicted.
Thus, in 2026, renewables and battery storage will account for all of net new utility-scale capacity additions.

EIA: 62% more renewable energy capacity is coming in 2026 electrek.co/2026/02/26/eia… #renewables #batteries #solar #wind

0 0 0 0

6 days ago

Text Shot: In ChatGPT, GPT‑5.4 Thinking can now provide an upfront plan of its thinking, so you can adjust course mid-response while it’s working, and arrive at a final output that’s more closely aligned with what you need without additional turns. GPT‑5.4 Thinking also improves deep web research, particularly for highly specific queries, while better maintaining context for questions that require longer thinking. Together, these improvements mean higher-quality answers that arrive faster and stay relevant to the task at hand.

Introducing GPT-5.4 openai.com/index/introduc… #AI #ChatGPT

2 0 0 1

6 days ago

Text Shot: Several thousand members of the campus community are already piloting the first tool, U of A Gen AI, which allows users to choose from five commercially available large language models: Claude, OpenAI, Gemma, Meta Llama and Amazon Nova.

In addition to U of A Gen AI, the platform will eventually include additional services to support teaching and learning, research and operations across campus. These include:

A public-facing campus chatbot that uses verified university information to answer questions about programs, services and resources
An Arizona AI assistant that will allow colleges, departments and offices to create custom chatbots that retrieve information only from approved institutional data sources
AI sandboxes that provide controlled, secure environments where researchers, instructors and students can test and experiment with AI systems without exposing sensitive or restricted data

Introducing the U of Arizona AI Platform news.arizona.edu/uannounce/intr… (our campus chatbots) #AI #education

6 1 0 1

6 days ago

Text Shot: The advent of gen AI has compressed the cycles of change quite dramatically. The accelerated movements of technology and abrupt turns in society mean that libraries need to be attentive and responsive to these changes. New tools and technologies are emerging with the potential to help libraries address the challenges and strengthen their position in society.

THE SYSTEMS LIBRARIAN - Anticipating Trends for 2026 and Beyond www.infotoday.com/cilmag/mar26/B… #AI #libraries #trends #change

1 1 3 0

6 days ago

Text Shot: A Claude Cowork plugin that performs comprehensive WCAG 2.2 Level AA accessibility audits on web pages and HTML files, with programmatic contrast verification and professional audit reports.

GitHub - CFLW-AI/wcag-audit-claude-skill github.com/CFLW-AI/wcag-a… #AI #accessibility #ClaudeCowork

3 0 0 0

6 days ago

Text Shot: Today I am sharing the Agent Protocols Tech Tree. APTT is a visual, videogame-style tech tree of the evolving protocols supporting AI agents.
Where did this come from?
I made the APTT for a session on “The Role of Protocols in the Agents Ecosystem” at the Towards an Internet Ecosystem for Sane Autonomous Agents workshop at the Berkman Klein Center on February 9th.
It’s a video game tech tree because, while the word “protocols” is boring, the phenomenon of open protocols is fascinating, and I want to make them easier to approach and explore.

Launching the Agent Protocols Tech Tree | Library Innovation Lab lil.law.harvard.edu/blog/2026/02/2… #AI #agents #protocols #learning

3 0 0 0

6 days ago

Creative Work Is About to Look a Lot More Like Programming every.to/thesis/creativ… #AI #design #images #video #systems

3 0 0 1

6 days ago

The next capability this unlocks: once you have a weekly mental model running, you can start using the synthesis output as context for decision-making prompts,
"Given what I now understand about [topic], help me think through [decision]."
That's where an LLM stops being a news reader and starts being an actual thinking partner. That's the next post.
Dedicated to all news agencies who lied about Puerto Vallarta, and continue to lie about Mexico. The people suffer because of your need for clicks. I see the suffering.

Your New HYPE FREE AI News Briefing System phillipalcock.substack.com/p/your-new-hyp… #AI #news #prompting

0 0 0 0

1 week ago

In effect, what will happen is that as AI makes it trivial to do some tasks, we will invest more in other tasks. The time to complete a paper will not change and the resources to do so will rise. We will likely continue the trend towards multi-
authored papers and labs like the Chetty one but the point is that there will be an escalation in effort that will come alongside the new thresholds for acceptance at top journals. Put simply, there will be no vibe researching your way into that new club.

Finally, I want to harken back to something I discussed when I discovered Al's ability to generate papers a year ago.
Not only will the production function of' papers change but the demand for papers will change too. New technology has a way of causing us to really think about why we are doing what we are doing and Al is no exception.

Journopoclypse! -Joshua Gans doesn’t think so joshuagans.substack.com/p/journopoclyp… #AI #ScholarlyPublishing

0 0 1 0

1 week ago

Text Shot: ABSTRACT
Using AI glasses as a focal case, we examine challenges wearable technology pose to teaching, assessment, and academic integrity. We argue that treating AI glasses as cheating devices overlooks their potential for accessibility and cognitive support while failing to address the structural disruption these technologies bring to education. Current enforcement models to detect academic misconduct rely on observable behaviours and verifiable evidence, at least on the balance of probability. The discreteness of AI glasses, combined with increasing prevalence of wearable technology for medical use, disrupts these established enforcement models, while simultaneously increasing the risk of unequal scrutiny and procedural injustice. We argue that a shift away from technological prohibition toward rethinking assessment and pedagogical design and policy reform may be more constructive and learner centred. In a postplagiarism era, academic integrity can be sustained by…

AI Smart Glasses and the Future of Academic Integrity in a Postplagiarism Era | Canadian Perspectives on Academic Integrity journalhosting.ucalgary.ca/index.php/ai/a… #AI #education #SmartGlasses #accessibility

1 0 0 1

1 week ago

Anthropic, which has done more than any other Western AI company to frame China as a threat, may now be deemed the same
"supply chain risk" designation historically reserved for Chinese companies like Huawei. The mockery lands harder given that just weeks earlier, Anthropic was being called "Al Thanos" ("AIXER") after its February product releases wiped out software stocks (IBM down 13%, CrowdStrike down 6.5%).
But there's a second level of political irony.
The US government, which built its entire
AI export control regime around the premise that democracies develop AI differently from autocracies, spent this week threatening a company with criminal prosecution for refusing to enable domestic mass surveillance and fully autonomous weapons, the exact use cases Washington spent years warning China would pursue.

China Reacts to Anthropic-DoW www.chinatalk.media/p/china-reacts… #AI #China #Anthropic

8 0 1 0