Cultivating a responsible innovation mindset among future tech leaders — from timeshighereducation.com by Andreas Alexiou from the University of Southampton
The classroom is a perfect place to discuss the messy, real-world consequences of technological discoveries, writes Andreas Alexiou. Beyond ‘How?’, students should be asking ‘Should we…?’ and ‘What if…?’ questions around ethics and responsibility

University educators play a crucial role in guiding students to think about the next big invention and its implications for privacy, the environment and social equity. To truly make a difference, we need to bring ethics and responsibility into the classroom in a way that resonates with students. Here’s how.

Debating with industry pioneers on incorporating ethical frameworks in innovation, product development or technology adoption is eye-opening because it can lead to students confronting assumptions they hadn’t questioned before.

Students need more than just skills; they need a mindset that sticks with them long after graduation. By making ethics and responsibility a key part of the learning process, educators are doing more than preparing students for a career; they’re preparing them to navigate a world shaped by their choices.

 

Google I/O 2025: From research to reality — from blog.google
Here’s how we’re making AI more helpful with Gemini.


Google I/O 2025 LIVE — all the details about Android XR smart glasses, AI Mode, Veo 3, Gemini, Google Beam and more — from tomsguide.com by Philip Michaels
Google’s annual conference goes all in on AI

With a running time of 2 hours, Google I/O 2025 leaned heavily into Gemini and new models that make the assistant work in more places than ever before. Despite focusing the majority of the keynote around Gemini, Google saved its most ambitious and anticipated announcement towards the end with its big Android XR smart glasses reveal.

Shockingly, very little was spent around Android 16. Most of its Android 16 related news, like the redesigned Material 3 Expressive interface, was announced during the Android Show live stream last week — which explains why Google I/O 2025 was such an AI heavy showcase.

That’s because Google carved out most of the keynote to dive deeper into Gemini, its new models, and integrations with other Google services. There’s clearly a lot to unpack, so here’s all the biggest Google I/O 2025 announcements.


Our vision for building a universal AI assistant— from blog.google
We’re extending Gemini to become a world model that can make plans and imagine new experiences by simulating aspects of the world.

Making Gemini a world model is a critical step in developing a new, more general and more useful kind of AI — a universal AI assistant. This is an AI that’s intelligent, understands the context you are in, and that can plan and take action on your behalf, across any device.

By applying LearnLM capabilities, and directly incorporating feedback from experts across the industry, Gemini adheres to the principles of learning science to go beyond just giving you the answer. Instead, Gemini can explain how you get there, helping you untangle even the most complex questions and topics so you can learn more effectively. Our new prompting guide provides sample instructions to see this in action.


Learn in newer, deeper ways with Gemini — from blog.google.com by Ben Gomes
We’re infusing LearnLM directly into Gemini 2.5 — plus more learning news from I/O.

At I/O 2025, we announced that we’re infusing LearnLM directly into Gemini 2.5, which is now the world’s leading model for learning. As detailed in our latest report, Gemini 2.5 Pro outperformed competitors on every category of learning science principles. Educators and pedagogy experts preferred Gemini 2.5 Pro over other offerings across a range of learning scenarios, both for supporting a user’s learning goals and on key principles of good pedagogy.


Gemini gets more personal, proactive and powerful — from blog.google.com by Josh Woodward
It’s your turn to create, learn and explore with an AI assistant that’s starting to understand your world and anticipate your needs.

Here’s what we announced at Google IO:

  • Gemini Live with camera and screen sharing, is now free on Android and iOS for everyone, so you can point your phone at anything and talk it through.
  • Imagen 4, our new image generation model, comes built in and is known for its image quality, better text rendering and speed.
  • Veo 3, our new, state-of-the-art video generation model, comes built in and is the first in the world to have native support for sound effects, background noises and dialogue between characters.
  • Deep Research and Canvas are getting their biggest updates yet, unlocking new ways to analyze information, create podcasts and vibe code websites and apps.
  • Gemini is coming to Chrome, so you can ask questions while browsing the web.
  • Students around the world can easily make interactive quizzes, and college students in the U.S., Brazil, Indonesia, Japan and the UK are eligible for a free school year of the Google AI Pro plan.
  • Google AI Ultra, a new premium plan, is for the pioneers who want the highest rate limits and early access to new features in the Gemini app.
  • 2.5 Flash has become our new default model, and it blends incredible quality with lightning fast response times.

Fuel your creativity with new generative media models and tools — from by Eli Collins
Introducing Veo 3 and Imagen 4, and a new tool for filmmaking called Flow.


AI in Search: Going beyond information to intelligence
We’re introducing new AI features to make it easier to ask any question in Search.

AI in Search is making it easier to ask Google anything and get a helpful response, with links to the web. That’s why AI Overviews is one of the most successful launches in Search in the past decade. As people use AI Overviews, we see they’re happier with their results, and they search more often. In our biggest markets like the U.S. and India, AI Overviews is driving over 10% increase in usage of Google for the types of queries that show AI Overviews.

This means that once people use AI Overviews, they’re coming to do more of these types of queries, and what’s particularly exciting is how this growth increases over time. And we’re delivering this at the speed people expect of Google Search — AI Overviews delivers the fastest AI responses in the industry.

In this story:

  • AI Mode in Search
  • Deep Search
  • Live capabilities
  • Agentic capabilities
  • Shopping
  • Personal context
  • Custom charts

 

 

I’m a LinkedIn Executive. I See the Bottom Rung of the Career Ladder Breaking. — from nytimes.com by Aneesh Raman; this is a gifted article

There are growing signs that artificial intelligence poses a real threat to a substantial number of the jobs that normally serve as the first step for each new generation of young workers. Uncertainty around tariffs and global trade is likely to only accelerate that pressure, just as millions of 2025 graduates enter the work force.

Breaking first is the bottom rung of the career ladder. In tech, advanced coding tools are creeping into the tasks of writing simple code and debugging — the ways junior developers gain experience. In law firms, junior paralegals and first-year associates who once cut their teeth on document review are handing weeks of work over to A.I. tools to complete in a matter of hours. And across retailers, A.I. chatbots and automated customer service tools are taking on duties once assigned to young associates.

 

.
Building an AI-Ready Workforce: A look at College Student ChatGPT Adoption in the US — from cdn.openai.com

One finding from our student survey that stood out to us: Many college and university students are teaching themselves and their friends about AI without waiting for their institutions to provide formal AI education or clear policies about the technology’s use. The education ecosystem is in an important moment of exploration and learning, but the rapid adoption by students across the country who haven’t received formalized instruction in how and when to use the technology creates disparities in AI access and knowledge.

The enclosed snapshot of how young people are using ChatGPT provides insight into the state of AI use among America’s college-aged students. We also include actionable proposals to help address adoption gaps. We hope these insights and proposals can inform research and policy conversation across the nation’s education ecosystem about how to achieve outcomes that support our students, our workforce, and the economy. By improving literacy, expanding access, and implementing clear policies, policymakers and educators can better integrate AI into our educational infrastructure and ensure that our workforce is ready to both sustain and benefit from our future with AI.

Leah Belsky | Vice President, Education | OpenAI

 

Top student use cases of ChatGPT -- learning and tutoring, writing help, miscellaneouc questions, and programming help

 
 

1-800-CHAT-GPT—12 Days of OpenAI: Day 10

Per The Rundown: OpenAI just launched a surprising new way to access ChatGPT — through an old-school 1-800 number & also rolled out a new WhatsApp integration for global users during Day 10 of the company’s livestream event.


How Agentic AI is Revolutionizing Customer Service — from customerthink.com by Devashish Mamgain

Agentic AI represents a significant evolution in artificial intelligence, offering enhanced autonomy and decision-making capabilities beyond traditional AI systems. Unlike conventional AI, which requires human instructions, agentic AI can independently perform complex tasks, adapt to changing environments, and pursue goals with minimal human intervention.

This makes it a powerful tool across various industries, especially in the customer service function. To understand it better, let’s compare AI Agents with non-AI agents.

Characteristics of Agentic AI

    • Autonomy: Achieves complex objectives without requiring human collaboration.
    • Language Comprehension: Understands nuanced human speech and text effectively.
    • Rationality: Makes informed, contextual decisions using advanced reasoning engines.
    • Adaptation: Adjusts plans and goals in dynamic situations.
    • Workflow Optimization: Streamlines and organizes business workflows with minimal oversight.

Clio: A system for privacy-preserving insights into real-world AI use — from anthropic.com

How, then, can we research and observe how our systems are used while rigorously maintaining user privacy?

Claude insights and observations, or “Clio,” is our attempt to answer this question. Clio is an automated analysis tool that enables privacy-preserving analysis of real-world language model use. It gives us insights into the day-to-day uses of claude.ai in a way that’s analogous to tools like Google Trends. It’s also already helping us improve our safety measures. In this post—which accompanies a full research paper—we describe Clio and some of its initial results.


Evolving tools redefine AI video — from heatherbcooper.substack.com by Heather Cooper
Google’s Veo 2, Kling 1.6, Pika 2.0 & more

AI video continues to surpass expectations
The AI video generation space has evolved dramatically in recent weeks, with several major players introducing groundbreaking tools.

Here’s a comprehensive look at the current landscape:

  • Veo 2…
  • Pika 2.0…
  • Runway’s Gen-3…
  • Luma AI Dream Machine…
  • Hailuo’s MiniMax…
  • OpenAI’s Sora…
  • Hunyuan Video by Tencent…

There are several other video models and platforms, including …

 

Best of 2024 — from wondertools.substack.com by Jeremy Caplan
12 of my favorites this year

I tested hundreds of new tools this year. Many were duplicative. A few stuck with me because they’re so useful. The dozen noted below are helping me mine insights from notes, summarize meetings, design visuals— even code a little, without being a developer. You can start using any of these in minutes — no big budget or prompt engineering PhD required.

 

Where to start with AI agents: An introduction for COOs — from fortune.com by Ganesh Ayyar

Picture your enterprise as a living ecosystem, where surging market demand instantly informs staffing decisions, where a new vendor’s onboarding optimizes your emissions metrics, where rising customer engagement reveals product opportunities. Now imagine if your systems could see these connections too! This is the promise of AI agents — an intelligent network that thinks, learns, and works across your entire enterprise.

Today, organizations operate in artificial silos. Tomorrow, they could be fluid and responsive. The transformation has already begun. The question is: will your company lead it?

The journey to agent-enabled operations starts with clarity on business objectives. Leaders should begin by mapping their business’s critical processes. The most pressing opportunities often lie where cross-functional handoffs create friction or where high-value activities are slowed by system fragmentation. These pain points become the natural starting points for your agent deployment strategy.


Create podcasts in minutes — from elevenlabs.io by Eleven Labs
Now anyone can be a podcast producer


Top AI tools for business — from theneuron.ai


This week in AI: 3D from images, video tools, and more — from heatherbcooper.substack.com by Heather Cooper
From 3D worlds to consistent characters, explore this week’s AI trends

Another busy AI news week, so I organized it into categories:

  • Image to 3D
  • AI Video
  • AI Image Models & Tools
  • AI Assistants / LLMs
  • AI Creative Workflow: Luma AI Boards

Want to speak Italian? Microsoft AI can make it sound like you do. — this is a gifted article from The Washington Post;
A new AI-powered interpreter is expected to simulate speakers’ voices in different languages during Microsoft Teams meetings.

Artificial intelligence has already proved that it can sound like a human, impersonate individuals and even produce recordings of someone speaking different languages. Now, a new feature from Microsoft will allow video meeting attendees to hear speakers “talk” in a different language with help from AI.


What Is Agentic AI?  — from blogs.nvidia.com by Erik Pounds
Agentic AI uses sophisticated reasoning and iterative planning to autonomously solve complex, multi-step problems.

The next frontier of artificial intelligence is agentic AI, which uses sophisticated reasoning and iterative planning to autonomously solve complex, multi-step problems. And it’s set to enhance productivity and operations across industries.

Agentic AI systems ingest vast amounts of data from multiple sources to independently analyze challenges, develop strategies and execute tasks like supply chain optimization, cybersecurity vulnerability analysis and helping doctors with time-consuming tasks.


 

Closing the digital use divide with active and engaging learning — from eschoolnews.com by Laura Ascione
Students offered insight into how to use active learning, with digital tools, to boost their engagement

When it comes to classroom edtech use, digital tools have a drastically different impact when they are used actively instead of passively–a critical difference examined in the 2023-2024 Speak Up Research by Project Tomorrow.

Students also outlined their ideal active learning technologies:

  • Collaboration tools to support projects
  • Student-teacher communication tools
  • Online databases for self-directed research
  • Multi-media tools for creating new content
  • Online and digital games
  • AI tools to support personalized learning
  • Coding and computer programming resources
  • Online animations, simulations, and virtual labs
  • Virtual reality equipment and content
 

2024: The State of Generative AI in the Enterprise — from menlovc.com (Menlo Ventures)
The enterprise AI landscape is being rewritten in real time. As pilots give way to production, we surveyed 600 U.S. enterprise IT decision-makers to reveal the emerging winners and losers.

This spike in spending reflects a wave of organizational optimism; 72% of decision-makers anticipate broader adoption of generative AI tools in the near future. This confidence isn’t just speculative—generative AI tools are already deeply embedded in the daily work of professionals, from programmers to healthcare providers.

Despite this positive outlook and increasing investment, many decision-makers are still figuring out what will and won’t work for their businesses. More than a third of our survey respondents do not have a clear vision for how generative AI will be implemented across their organizations. This doesn’t mean they’re investing without direction; it simply underscores that we’re still in the early stages of a large-scale transformation. Enterprise leaders are just beginning to grasp the profound impact generative AI will have on their organizations.


Business spending on AI surged 500% this year to $13.8 billion, says Menlo Ventures — from cnbc.com by Hayden Field

Key Points

  • Business spending on generative AI surged 500% this year, hitting $13.8 billion — up from just $2.3 billion in 2023, according to data from Menlo Ventures released Wednesday.
  • OpenAI ceded market share in enterprise AI, declining from 50% to 34%, per the report.
  • Amazon-backed Anthropic doubled its market share from 12% to 24%.

Microsoft quietly assembles the largest AI agent ecosystem—and no one else is close — from venturebeat.com by Matt Marshall

Microsoft has quietly built the largest enterprise AI agent ecosystem, with over 100,000 organizations creating or editing AI agents through its Copilot Studio since launch – a milestone that positions the company ahead in one of enterprise tech’s most closely watched and exciting  segments.

The rapid adoption comes as Microsoft significantly expands its agent capabilities. At its Ignite conference [that started on 11/19/24], the company announced it will allow enterprises to use any of the 1,800 large language models (LLMs) in the Azure catalog within these agents – a significant move beyond its exclusive reliance on OpenAI’s models. The company also unveiled autonomous agents that can work independently, detecting events and orchestrating complex workflows with minimal human oversight.


Now Hear This: World’s Most Flexible Sound Machine Debuts — from
Using text and audio as inputs, a new generative AI model from NVIDIA can create any combination of music, voices and sounds.

Along these lines, also see:


AI Agents Versus Human Agency: 4 Ways To Navigate Our AI-Driven World — from forbes.com by Cornelia C. Walther

To understand the implications of AI agents, it’s useful to clarify the distinctions between AI, generative AI, and AI agents and explore the opportunities and risks they present to our autonomy, relationships, and decision-making.

AI Agents: These are specialized applications of AI designed to perform tasks or simulate interactions. AI agents can be categorized into:

    • Tool Agents…
    • Simulation Agents..

While generative AI creates outputs from prompts, AI agents use AI to act with intention, whether to assist (tool agents) or emulate (simulation agents). The latter’s ability to mirror human thought and action offers fascinating possibilities — and raises significant risks.

 

2024-11-22: The Race to the TopDario Amodei on AGI, Risks, and the Future of Anthropic — from emergentbehavior.co by Prakash (Ate-a-Pi)

Risks on the Horizon: ASL Levels
The two key risks Dario is concerned about are:

a) cyber, bio, radiological, nuclear (CBRN)
b) model autonomy

These risks are captured in Anthropic’s framework for understanding AI Safety Levels (ASL):

1. ASL-1: Narrow-task AI like Deep Blue (no autonomy, minimal risk).
2. ASL-2: Current systems like ChatGPT/Claude, which lack autonomy and don’t pose significant risks beyond information already accessible via search engines.
3. ASL-3: Agents arriving soon (potentially next year) that can meaningfully assist non-state actors in dangerous activities like cyber or CBRN (chemical, biological, radiological, nuclear) attacks. Security and filtering are critical at this stage to prevent misuse.
4. ASL-4: AI smart enough to evade detection, deceive testers, and assist state actors with dangerous projects. AI will be strong enough that you would want to use the model to do anything dangerous. Mechanistic interpretability becomes crucial for verifying AI behavior.
5. ASL-5: AGI surpassing human intelligence in all domains, posing unprecedented challenges.

Anthropic’s if/then framework ensures proactive responses: if a model demonstrates danger, the team clamps down hard, enforcing strict controls.



Should You Still Learn to Code in an A.I. World? — from nytimes.com by
Coding boot camps once looked like the golden ticket to an economically secure future. But as that promise fades, what should you do? Keep learning, until further notice.

Compared with five years ago, the number of active job postings for software developers has dropped 56 percent, according to data compiled by CompTIA. For inexperienced developers, the plunge is an even worse 67 percent.
“I would say this is the worst environment for entry-level jobs in tech, period, that I’ve seen in 25 years,” said Venky Ganesan, a partner at the venture capital firm Menlo Ventures.

For years, the career advice from everyone who mattered — the Apple chief executive Tim Cook, your mother — was “learn to code.” It felt like an immutable equation: Coding skills + hard work = job.

Now the math doesn’t look so simple.

Also see:

AI builds apps in 2 mins flat — where the Neuron mentions this excerpt about Lovable:

There’s a new coding startup in town, and it just MIGHT have everybody else shaking in their boots (we’ll qualify that in a sec, don’t worry).

It’s called Lovable, the “world’s first AI fullstack engineer.”

Lovable does all of that by itself. Tell it what you want to build in plain English, and it creates everything you need. Want users to be able to log in? One click. Need to store data? One click. Want to accept payments? You get the idea.

Early users are backing up these claims. One person even launched a startup that made Product Hunt’s top 10 using just Lovable.

As for us, we made a Wordle clone in 2 minutes with one prompt. Only edit needed? More words in the dictionary. It’s like, really easy y’all.


When to chat with AI (and when to let it work) — from aiwithallie.beehiiv.com by Allie K. Miller

Re: some ideas on how to use Notebook LM:

  • Turn your company’s annual report into an engaging podcast
  • Create an interactive FAQ for your product manual
  • Generate a timeline of your industry’s history from multiple sources
  • Produce a study guide for your online course content
  • Develop a Q&A system for your company’s knowledge base
  • Synthesize research papers into digestible summaries
  • Create an executive content briefing from multiple competitor blog posts
  • Generate a podcast discussing the key points of a long-form research paper

Introducing conversation practice: AI-powered simulations to build soft skills — from codesignal.com by Albert Sahakyan

From DSC:
I have to admit I’m a bit suspicious here, as the “conversation practice” product seems a bit too scripted at times, but I post it because the idea of using AI to practice soft skills development makes a great deal of sense:


 

Miscommunication Leads AI-Based Hiring Tools Astray — from adigaskell.org

Nearly every Fortune 500 company now uses artificial intelligence (AI) to screen resumes and assess test scores to find the best talent. However, new research from the University of Florida suggests these AI tools might not be delivering the results hiring managers expect.

The problem stems from a simple miscommunication between humans and machines: AI thinks it’s picking someone to hire, but hiring managers only want a list of candidates to interview.

Without knowing about this next step, the AI might choose safe candidates. But if it knows there will be another round of screening, it might suggest different and potentially stronger candidates.


AI agents explained: Why OpenAI, Google and Microsoft are building smarter AI agents — from digit.in by Jayesh Shinde

In the last two years, the world has seen a lot of breakneck advancement in the Generative AI space, right from text-to-text, text-to-image and text-to-video based Generative AI capabilities. And all of that’s been nothing short of stepping stones for the next big AI breakthrough – AI agents. According to Bloomberg, OpenAI is preparing to launch its first autonomous AI agent, which is codenamed ‘Operator,’ as soon as in January 2025.

Apparently, this OpenAI agent – or Operator, as it’s codenamed – is designed to perform complex tasks independently. By understanding user commands through voice or text, this AI agent will seemingly do tasks related to controlling different applications in the computer, send an email, book flights, and no doubt other cool things. Stuff that ChatGPT, Copilot, Google Gemini or any other LLM-based chatbot just can’t do on its own.


2025: The year ‘invisible’ AI agents will integrate into enterprise hierarchies  — from venturebeat.com by Taryn Plumb

In the enterprise of the future, human workers are expected to work closely alongside sophisticated teams of AI agents.

According to McKinsey, generative AI and other technologies have the potential to automate 60 to 70% of employees’ work. And, already, an estimated one-third of American workers are using AI in the workplace — oftentimes unbeknownst to their employers.

However, experts predict that 2025 will be the year that these so-called “invisible” AI agents begin to come out of the shadows and take more of an active role in enterprise operations.

“Agents will likely fit into enterprise workflows much like specialized members of any given team,” said Naveen Rao, VP of AI at Databricks and founder and former CEO of MosaicAI.


State of AI Report 2024 Summary — from ai-supremacy.com by Michael Spencer
Part I, Consolidation, emergence and adoption. 


Which AI Image Model Is the Best Speller? Let’s Find Out! — from whytryai.com by Daniel Nest
I test 7 image models to find those that can actually write.

The contestants
I picked 7 participants for today’s challenge:

  1. DALL-E 3 by OpenAI (via Microsoft Designer)
  2. FLUX1.1 [pro] by Black Forest Labs (via Glif)
  3. Ideogram 2.0 by Ideogram (via Ideogram)
  4. Imagen 3 by Google (via Image FX)
  5. Midjourney 6.1 by Midjourney (via Midjourney)
  6. Recraft V3 by Recraft (via Recraft)
  7. Stable Diffusion 3.5 Large by Stability AI (via Hugging Face)

How to get started with AI agents (and do it right) — from venturebeat.com by Taryn Plumb

So how can enterprises choose when to adopt third-party models, open source tools or build custom, in-house fine-tuned models? Experts weigh in.


OpenAI, Google and Anthropic Are Struggling to Build More Advanced AI — from bloomberg.com (behind firewall)
Three of the leading artificial intelligence companies are seeing diminishing returns from their costly efforts to develop newer models.


OpenAI and others seek new path to smarter AI as current methods hit limitations — from reuters.com by Krystal Hu and Anna Tong

Summary

  • AI companies face delays and challenges with training new large language models
  • Some researchers are focusing on more time for inference in new models
  • Shift could impact AI arms race for resources like chips and energy

NVIDIA Advances Robot Learning and Humanoid Development With New AI and Simulation Tools — from blogs.nvidia.com by Spencer Huang
New Project GR00T workflows and AI world model development technologies to accelerate robot dexterity, control, manipulation and mobility.


How Generative AI is Revolutionizing Product Development — from intelligenthq.com

A recent report from McKinsey predicts that generative AI could unlock up to $2.6 to $4.4 annually trillion in value within product development and innovation across various industries. This staggering figure highlights just how significantly generative AI is set to transform the landscape of product development. Generative AI app development is driving innovation by using the power of advanced algorithms to generate new ideas, optimize designs, and personalize products at scale. It is also becoming a cornerstone of competitive advantage in today’s fast-paced market. As businesses look to stay ahead, understanding and integrating technologies like generative AI app development into product development processes is becoming more crucial than ever.


What are AI Agents: How To Create a Based AI Agent — from ccn.com by Lorena Nessi

Key Takeaways

  • AI agents handle complex, autonomous tasks beyond simple commands, showcasing advanced decision-making and adaptability.
  • The Based AI Agent template by Coinbase and Replit provides an easy starting point for developers to build blockchain-enabled AI agents.
  • AI based agents specifically integrate with blockchain, supporting crypto wallets and transactions.
  • Securing API keys in development is crucial to protect the agent from unauthorized access.

What are AI Agents and How Are They Used in Different Industries? — from rtinsights.com by Salvatore Salamone
AI agents enable companies to make smarter, faster, and more informed decisions. From predictive maintenance to real-time process optimization, these agents are delivering tangible benefits across industries.

 

Along these same lines, see:

Introducing computer use, a new Claude 3.5 Sonnet, and Claude 3.5 Haiku

We’re also introducing a groundbreaking new capability in public beta: computer use. Available today on the API, developers can direct Claude to use computers the way people do—by looking at a screen, moving a cursor, clicking buttons, and typing text. Claude 3.5 Sonnet is the first frontier AI model to offer computer use in public beta. At this stage, it is still experimental—at times cumbersome and error-prone. We’re releasing computer use early for feedback from developers, and expect the capability to improve rapidly over time.


ZombAIs: From Prompt Injection to C2 with Claude Computer Use — from embracethered.com by Johann Rehberger

A few days ago, Anthropic released Claude Computer Use, which is a model + code that allows Claude to control a computer. It takes screenshots to make decisions, can run bash commands and so forth.

It’s cool, but obviously very dangerous because of prompt injection. Claude Computer Use enables AI to run commands on machines autonomously, posing severe risks if exploited via prompt injection.

This blog post demonstrates that it’s possible to leverage prompt injection to achieve, old school, command and control (C2) when giving novel AI systems access to computers.

We discussed one way to get malware onto a Claude Computer Use host via prompt injection. There are countless others, like another way is to have Claude write the malware from scratch and compile it. Yes, it can write C code, compile and run it. There are many other options.

TrustNoAI.

And again, remember do not run unauthorized code on systems that you do not own or are authorized to operate on.

Also relevant here, see:


Perplexity Grows, GPT Traffic Surges, Gamma Dominates AI Presentations – The AI for Work Top 100: October 2024 — from flexos.work by Daan van Rossum
Perplexity continues to gain users despite recent controversies. Five out of six GPTs see traffic boosts. This month’s highest gainers including Gamma, Blackbox, Runway, and more.


Growing Up: Navigating Generative AI’s Early Years – AI Adoption Report — from ai.wharton.upenn.edu by  Jeremy Korst, Stefano Puntoni, & Mary Purk

From a survey with more than 800 senior business leaders, this report’s findings indicate that weekly usage of Gen AI has nearly doubled from 37% in 2023 to 72% in 2024, with significant growth in previously slower-adopting departments like Marketing and HR. Despite this increased usage, businesses still face challenges in determining the full impact and ROI of Gen AI. Sentiment reports indicate leaders have shifted from feelings of “curiosity” and “amazement” to more positive sentiments like “pleased” and “excited,” and concerns about AI replacing jobs have softened. Participants were full-time employees working in large commercial organizations with 1,000 or more employees.


Apple study exposes deep cracks in LLMs’ “reasoning” capabilities — from arstechnica.com by Kyle Orland
Irrelevant red herrings lead to “catastrophic” failure of logical inference.

For a while now, companies like OpenAI and Google have been touting advanced “reasoning” capabilities as the next big step in their latest artificial intelligence models. Now, though, a new study from six Apple engineers shows that the mathematical “reasoning” displayed by advanced large language models can be extremely brittle and unreliable in the face of seemingly trivial changes to common benchmark problems.

The fragility highlighted in these new results helps support previous research suggesting that LLMs use of probabilistic pattern matching is missing the formal understanding of underlying concepts needed for truly reliable mathematical reasoning capabilities. “Current LLMs are not capable of genuine logical reasoning,” the researchers hypothesize based on these results. “Instead, they attempt to replicate the reasoning steps observed in their training data.”


Google CEO says more than a quarter of the company’s new code is created by AI — from businessinsider.in by Hugh Langley

  • More than a quarter of new code at Google is made by AI and then checked by employees.
  • Google is doubling down on AI internally to make its business more efficient.

Top Generative AI Chatbots by Market Share – October 2024 


Bringing developer choice to Copilot with Anthropic’s Claude 3.5 Sonnet, Google’s Gemini 1.5 Pro, and OpenAI’s o1-preview — from github.blog

We are bringing developer choice to GitHub Copilot with Anthropic’s Claude 3.5 Sonnet, Google’s Gemini 1.5 Pro, and OpenAI’s o1-preview and o1-mini. These new models will be rolling out—first in Copilot Chat, with OpenAI o1-preview and o1-mini available now, Claude 3.5 Sonnet rolling out progressively over the next week, and Google’s Gemini 1.5 Pro in the coming weeks. From Copilot Workspace to multi-file editing to code review, security autofix, and the CLI, we will bring multi-model choice across many of GitHub Copilot’s surface areas and functions soon.

 

AI-governed robots can easily be hacked — from theaivalley.com by Barsee
PLUS: Sam Altman’s new company “World” introduced…

In a groundbreaking study, researchers from Penn Engineering showed how AI-powered robots can be manipulated to ignore safety protocols, allowing them to perform harmful actions despite normally rejecting dangerous task requests.

What did they find ?

  • Researchers found previously unknown security vulnerabilities in AI-governed robots and are working to address these issues to ensure the safe use of large language models(LLMs) in robotics.
  • Their newly developed algorithm, RoboPAIR, reportedly achieved a 100% jailbreak rate by bypassing the safety protocols on three different AI robotic systems in a few days.
  • Using RoboPAIR, researchers were able to manipulate test robots into performing harmful actions, like bomb detonation and blocking emergency exits, simply by changing how they phrased their commands.

Why does it matter?

This research highlights the importance of spotting weaknesses in AI systems to improve their safety, allowing us to test and train them to prevent potential harm.

From DSC:
Great! Just what we wanted to hear. But does it surprise anyone? Even so…we move forward at warp speeds.


From DSC:
So, given the above item, does the next item make you a bit nervous as well? I saw someone on Twitter/X exclaim, “What could go wrong?”  I can’t say I didn’t feel the same way.

Introducing computer use, a new Claude 3.5 Sonnet, and Claude 3.5 Haiku — from anthropic.com

We’re also introducing a groundbreaking new capability in public beta: computer use. Available today on the API, developers can direct Claude to use computers the way people do—by looking at a screen, moving a cursor, clicking buttons, and typing text. Claude 3.5 Sonnet is the first frontier AI model to offer computer use in public beta. At this stage, it is still experimental—at times cumbersome and error-prone. We’re releasing computer use early for feedback from developers, and expect the capability to improve rapidly over time.

Per The Rundown AI:

The Rundown: Anthropic just introduced a new capability called ‘computer use’, alongside upgraded versions of its AI models, which enables Claude to interact with computers by viewing screens, typing, moving cursors, and executing commands.

Why it matters: While many hoped for Opus 3.5, Anthropic’s Sonnet and Haiku upgrades pack a serious punch. Plus, with the new computer use embedded right into its foundation models, Anthropic just sent a warning shot to tons of automation startups—even if the capabilities aren’t earth-shattering… yet.

Also related/see:

  • What is Anthropic’s AI Computer Use? — from ai-supremacy.com by Michael Spencer
    Task automation, AI at the intersection of coding and AI agents take on new frenzied importance heading into 2025 for the commercialization of Generative AI.
  • New Claude, Who Dis? — from theneurondaily.com
    Anthropic just dropped two new Claude models…oh, and Claude can now use your computer.
  • When you give a Claude a mouse — from oneusefulthing.org by Ethan Mollick
    Some quick impressions of an actual agent

Introducing Act-One — from runwayml.com
A new way to generate expressive character performances using simple video inputs.

Per Lore by Nathan Lands:

What makes Act-One special? It can capture the soul of an actor’s performance using nothing but a simple video recording. No fancy motion capture equipment, no complex face rigging, no army of animators required. Just point a camera at someone acting, and watch as their exact expressions, micro-movements, and emotional nuances get transferred to an AI-generated character.

Think about what this means for creators: you could shoot an entire movie with multiple characters using just one actor and a basic camera setup. The same performance can drive characters with completely different proportions and looks, while maintaining the authentic emotional delivery of the original performance. We’re witnessing the democratization of animation tools that used to require millions in budget and years of specialized training.

Also related/see:


Google to buy nuclear power for AI datacentres in ‘world first’ deal — from theguardian.com
Tech company orders six or seven small nuclear reactors from California’s Kairos Power

Google has signed a “world first” deal to buy energy from a fleet of mini nuclear reactors to generate the power needed for the rise in use of artificial intelligence.

The US tech corporation has ordered six or seven small nuclear reactors (SMRs) from California’s Kairos Power, with the first due to be completed by 2030 and the remainder by 2035.

Related:


ChatGPT Topped 3 Billion Visits in September — from similarweb.com

After the extreme peak and summer slump of 2023, ChatGPT has been setting new traffic highs since May

ChatGPT has been topping its web traffic records for months now, with September 2024 traffic up 112% year-over-year (YoY) to 3.1 billion visits, according to Similarweb estimates. That’s a change from last year, when traffic to the site went through a boom-and-bust cycle.


Crazy “AI Army” — from aisecret.us

Also from aisecret.us, see World’s First Nuclear Power Deal For AI Data Centers

Google has made a historic agreement to buy energy from a group of small nuclear reactors (SMRs) from Kairos Power in California. This is the first nuclear power deal specifically for AI data centers in the world.


New updates to help creators build community, drive business, & express creativity on YouTube — from support.google.com

Hey creators!
Made on YouTube 2024 is here and we’ve announced a lot of updates that aim to give everyone the opportunity to build engaging communities, drive sustainable businesses, and express creativity on our platform.

Below is a roundup with key info – feel free to upvote the announcements that you’re most excited about and subscribe to this post to get updates on these features! We’re looking forward to another year of innovating with our global community it’s a future full of opportunities, and it’s all Made on YouTube!


New autonomous agents scale your team like never before — from blogs.microsoft.com

Today, we’re announcing new agentic capabilities that will accelerate these gains and bring AI-first business process to every organization.

  • First, the ability to create autonomous agents with Copilot Studio will be in public preview next month.
  • Second, we’re introducing ten new autonomous agents in Dynamics 365 to build capacity for every sales, service, finance and supply chain team.

10 Daily AI Use Cases for Business Leaders— from flexos.work by Daan van Rossum
While AI is becoming more powerful by the day, business leaders still wonder why and where to apply today. I take you through 10 critical use cases where AI should take over your work or partner with you.


Multi-Modal AI: Video Creation Simplified — from heatherbcooper.substack.com by Heather Cooper

Emerging Multi-Modal AI Video Creation Platforms
The rise of multi-modal AI platforms has revolutionized content creation, allowing users to research, write, and generate images in one app. Now, a new wave of platforms is extending these capabilities to video creation and editing.

Multi-modal video platforms combine various AI tools for tasks like writing, transcription, text-to-voice conversion, image-to-video generation, and lip-syncing. These platforms leverage open-source models like FLUX and LivePortrait, along with APIs from services such as ElevenLabs, Luma AI, and Gen-3.


AI Medical Imagery Model Offers Fast, Cost-Efficient Expert Analysis — from developer.nvidia.com/

 

From DSC:
Great…we have another tool called Canvas. Or did you say Canva?

Introducing canvas — from OpenAI
A new way of working with ChatGPT to write and code

We’re introducing canvas, a new interface for working with ChatGPT on writing and coding projects that go beyond simple chat. Canvas opens in a separate window, allowing you and ChatGPT to collaborate on a project. This early beta introduces a new way of working together—not just through conversation, but by creating and refining ideas side by side.

Canvas was built with GPT-4o and can be manually selected in the model picker while in beta. Starting today we’re rolling out canvas to ChatGPT Plus and Team users globally. Enterprise and Edu users will get access next week. We also plan to make canvas available to all ChatGPT Free users when it’s out of beta.


Using AI to buy your home? These companies think it’s time you should — from usatoday.com by Andrea Riquier

The way Americans buy homes is changing dramatically.

New industry rules about how home buyers’ real estate agents get paid are prompting a reckoning among housing experts and the tech sector. Many house hunters who are already stretched thin by record-high home prices and closing costs must now decide whether, and how much, to pay an agent.

A 2-3% commission on the median home price of $416,700 could be well over $10,000, and in a world where consumers are accustomed to using technology for everything from taxes to tickets, many entrepreneurs see an opportunity to automate away the middleman, even as some consumer advocates say not so fast.


The State of AI Report 2024 — from nathanbenaich.substack.com by Nathan Benaich


The Great Mismatch — from the-job.beehiiv.com. by Paul Fain
Artificial intelligence could threaten millions of decent-paying jobs held by women without degrees.

Women in administrative and office roles may face the biggest AI automation risk, find Brookings researchers armed with data from OpenAI. Also, why Indiana could make the Swiss apprenticeship model work in this country, and how learners get disillusioned when a certificate doesn’t immediately lead to a good job.

major new analysis from the Brookings Institution, using OpenAI data, found that the most vulnerable workers don’t look like the rail and dockworkers who have recaptured the national spotlight. Nor are they the creatives—like Hollywood’s writers and actors—that many wealthier knowledge workers identify with. Rather, they’re predominantly women in the 19M office support and administrative jobs that make up the first rung of the middle class.

“Unfortunately the technology and automation risks facing women have been overlooked for a long time,” says Molly Kinder, a fellow at Brookings Metro and lead author of the new report. “Most of the popular and political attention to issues of automation and work centers on men in blue-collar roles. There is far less awareness about the (greater) risks to women in lower-middle-class roles.”



Is this how AI will transform the world over the next decade? — from futureofbeinghuman.com by Andrew Maynard
Anthropic’s CEO Dario Amodei has just published a radical vision of an AI-accelerated future. It’s audacious, compelling, and a must-read for anyone working at the intersection of AI and society.

But if Amodei’s essay is approached as a conversation starter rather than a manifesto — which I think it should be — it’s hard to see how it won’t lead to clearer thinking around how we successfully navigate the coming AI transition.

Given the scope of the paper, it’s hard to write a response to it that isn’t as long or longer as the original. Because of this, I’d strongly encourage anyone who’s looking at how AI might transform society to read the original — it’s well written, and easier to navigate than its length might suggest.

That said, I did want to pull out a few things that struck me as particularly relevant and important — especially within the context of navigating advanced technology transitions.

And speaking of that essay, here’s a summary from The Rundown AI:

Anthropic CEO Dario Amodei just published a lengthy essay outlining an optimistic vision for how AI could transform society within 5-10 years of achieving human-level capabilities, touching on longevity, politics, work, the economy, and more.

The details:

  • Amodei believes that by 2026, ‘powerful AI’ smarter than a Nobel Prize winner across fields, with agentic and all multimodal capabilities, will be possible.
  • He also predicted that AI could compress 100 years of scientific progress into 10 years, curing most diseases and doubling the human lifespan.
  • The essay argued AI could strengthen democracy by countering misinformation and providing tools to undermine authoritarian regimes.
  • The CEO acknowledged potential downsides, including job displacement — but believes new economic models will emerge to address this.
  • He envisions AI driving unprecedented economic growth but emphasizes ensuring AI’s benefits are broadly distributed.

Why it matters: 

  • As the CEO of what is seen as the ‘safety-focused’ AI lab, Amodei paints a utopia-level optimistic view of where AI will head over the next decade. This thought-provoking essay serves as both a roadmap for AI’s potential and a call to action to ensure the responsible development of technology.

AI in the Workplace: Answering 3 Big Questions — from gallup.com by Kate Den Houter

However, most workers remain unaware of these efforts. Only a third (33%) of all U.S. employees say their organization has begun integrating AI into their business practices, with the highest percentage in white-collar industries (44%).

White-collar workers are more likely to be using AI. White-collar workers are, by far, the most frequent users of AI in their roles. While 81% of employees in production/frontline industries say they never use AI, only 54% of white-collar workers say they never do and 15% report using AI weekly.

Most employees using AI use it for idea generation and task automation. Among employees who say they use AI, the most common uses are to generate ideas (41%), to consolidate information or data (39%), and to automate basic tasks (39%).


Nvidia Blackwell GPUs sold out for the next 12 months as AI market boom continues — from techspot.com by Skye Jacobs
Analysts expect Team Green to increase its already formidable market share

Selling like hotcakes: The extraordinary demand for Blackwell GPUs illustrates the need for robust, energy-efficient processors as companies race to implement more sophisticated AI models and applications. The coming months will be critical to Nvidia as the company works to ramp up production and meet the overwhelming requests for its latest product.


Here’s my AI toolkit — from wondertools.substack.com by Jeremy Caplan and Nikita Roy
How and why I use the AI tools I do — an audio conversation

1. What are two useful new ways to use AI?

  • AI-powered research: Type a detailed search query into Perplexity instead of Google to get a quick, actionable summary response with links to relevant information sources. Read more of my take on why Perplexity is so useful and how to use it.
  • Notes organization and analysis: Tools like NotebookLM, Claude Projects, and Mem can help you make sense of huge repositories of notes and documents. Query or summarize your own notes and surface novel connections between your ideas.
 
© 2025 | Daniel Christian