Another Crazy Day in AI 9/104

Another Crazy Day in AI: The Cognitive Cost of Convenience

Oct 17
4 min read

Hello, AI Enthusiasts.

Made it through the week? Perfect time for a read that stretches your thinking.

An independent researcher just released a deep study exploring how AI might already be changing how we process thoughts and make choices. It’s part science, part philosophy, and all very human.

Meanwhile, Google and PwC are thinking macro: how AI could help developing countries do more with less.

And to cap things off, Meta’s adding new parental controls so teens can talk to AIs safely (and parents can finally breathe a little easier).

Take the weekend to think — or not think — about it.

Here's another crazy day in AI:

When algorithms shape how we think
Google report shows how AI can power better public services
Meta introduces new tools to help parents manage teen AI use
Some AI tools to try out

TODAY'S FEATURED ITEM: Minds in the Age of Algorithms

A robotic scientist in a classic white coat with 'AI Scientist' on its back stands beside a human scientist with 'Human Scientist' on their coat, looking towards the AI Scientist.

Image Credit: Wowza (created with Ideogram)

How much of our thinking still comes from us, and how much do we now outsource?

Independent researcher Rénald Gesnot has published a 127-page technical monograph examining how artificial intelligence is reshaping human cognition, thought patterns, and decision-making. Released as a working paper in July 2025 on arXiv, this research offers a careful, multidimensional look at how AI is influencing the way we think, reason, and engage with the world. Drawing from cognitive science, ethics, and social analysis, Gesnot explores both the advantages and the concerns tied to a growing reliance on intelligent systems in daily life.

The study discusses several major areas:

Cognitive offloading — AI makes it easier to handle complex tasks, but frequent reliance on it may reduce our engagement in reasoning, problem-solving, and memory retention.
Cognitive standardization — With more people depending on the same algorithms and datasets, human expression and creativity may become more uniform over time.
Algorithmic influence — Personalization systems determine much of what we see and read, subtly influencing preferences, opinions, and the range of ideas we encounter.
Artificial consciousness — The paper considers whether machines could one day demonstrate self-awareness, raising questions about moral responsibility and consciousness itself.
Human responsibility — Gesnot emphasizes the importance of education, transparency, and strong governance to preserve independent thought and intellectual diversity.

Gesnot’s paper doesn’t read like a warning, but more like an invitation to think carefully about how technology is changing the way we process information. He reminds readers that every tool, no matter how intelligent, comes with trade-offs. While AI helps us reason faster and explore ideas at scale, it may also make it easier to accept quick answers instead of working through complex questions ourselves. It’s a subtle but meaningful shift in how we define “thinking” — not just as producing ideas, but as understanding how they’re formed.

At a broader level, the study urges reflection on how shared systems influence shared understanding. If the same algorithms guide what billions of people read, write, and believe, the diversity of human thought could gradually thin out. Not through censorship, but through repetition — where convenience and familiarity quietly take the place of curiosity and exploration. Gesnot’s work points out that keeping thought diverse isn’t just a creative concern; it’s essential for how societies grow, challenge themselves, and adapt to change.

In the end, the paper leaves us with a quiet reminder rather than a conclusion. AI will keep evolving — that much is certain. What remains uncertain is how consciously we’ll shape our relationship with it. The future of thinking may depend less on what machines can do, and more on how willing we are to stay curious, to question, and to keep thinking for ourselves even as technology learns to do more of it with us.

Read the full paper here.

OTHER INTERESTING AI HIGHLIGHTS:

Google Report Shows How AI Can Power Better Public Services

/Karan Bhatia, Global Head, Government Affairs & Public Policy, on Google Blogs - The Keyword

Google and PwC’s new AI Sprinters report outlines how developing countries can use AI to modernize public services, strengthen economies, and reduce fiscal burdens. The report identifies four readiness archetypes—from early “Explorers” to advanced “Leaders”—to help governments map their AI journeys. By leveraging AI to improve productivity, transparency, and inclusion, developing nations could reduce deficits by up to 22% and boost GDP by 4% by 2035. The findings highlight how AI can help governments leapfrog legacy systems and build stronger, more efficient institutions.

Read more here.

Meta Introduces New Tools To Help Parents Manage Teen AI use

/Adam Mosseri, Head of Instagram, and Alexandr Wang, Chief AI Officer, on Meta Newsroom

Meta is introducing new parental tools to help families manage how teens interact with AI on its platforms. Parents will soon be able to turn off one-on-one chats with AI characters, block specific AIs, and view insights into their teens’ conversations. These updates build on Meta’s existing PG-13 safety framework for AI responses, designed to keep interactions age-appropriate and educational. The company says its goal is to ensure AI complements learning and creativity while maintaining strong safeguards for teen users.

Read more here.

SOME AI TOOLS TO TRY OUT:

FLORA – AI canvas for brand teams to brainstorm, refine, and create with full control.
JustPaid – Automates invoicing, tracking, and revenue insights to simplify billing.
Strella – Runs AI-powered customer interviews and delivers insights in hours.

That’s a wrap on today’s Almost Daily craziness.

Catch us almost every day—almost! 😉

EXCITING NEWS:

The Another Crazy Day in AI newsletter is on LinkedIn!!!

Leveraging AI for Enhanced Content: As part of our commitment to exploring new technologies, we used AI to help curate and refine our newsletters. This enriches our content and keeps us at the forefront of digital innovation, ensuring you stay informed with the latest trends and developments.

Another Crazy Day in AI: Your Windows PC Now Understands What You See and Say

Oct 17
4 min read

Hello, AI Enthusiasts.

We’re crawling toward the weekend, but the AI world clearly didn’t get the memo.

Windows 11 is getting its biggest AI infusion yet, bringing Copilot’s smarts to every device. No more wrestling with perfect prompts — these updates make it as simple as talking, pointing, or asking out loud.

Meanwhile, Spotify is striking a rare harmony with major record labels to ensure AI helps artists, not replaces them.

And speaking of smarter systems, Anthropic just gave Claude an upgrade with new skills that help it handle specialized work more efficiently.

Here's another crazy day in AI:

Microsoft transforms PCs into conversational AI assistants
Spotify partners with record labels to create artist-first AI tools
Claude becomes more customizable with Skills
Some AI tools to try out

TODAY'S FEATURED ITEM: Windows 11 Gets Voice Activation and Vision AI

Image Credit: Wowza (created with Ideogram)

How much time do you spend telling your computer what to do versus having it actually understand what you need?

Microsoft has rolled out a comprehensive set of AI updates for Windows 11, bringing Copilot's capabilities to every device running the operating system. The announcement from Yusuf Mehdi, Executive Vice President and Consumer Chief Marketing Officer, outlines how the company is building AI functionality into everyday Windows experiences. Rather than requiring users to craft detailed prompts, these updates allow for more conversational interaction through voice and visual understanding. The changes represent Microsoft’s attempt to make AI tools accessible to people who might find traditional prompting tedious or unclear.

Among the updates introduced:

Voice activation through "Hey Copilot" enables hands-free interaction, with the company noting users engage twice as much through voice compared to typing
Copilot Vision rolls out globally, reading what's on your screen to offer guidance whether you're editing images, reviewing documents, or figuring out new software
Taskbar integration puts Copilot one click away, plus a refreshed search experience for finding apps, files, and settings faster
Copilot Actions enters preview, letting the AI handle background tasks like sorting vacation photos or pulling data from PDFs while you do other work
Connected services now include OneDrive, Outlook, Google Drive, Gmail, and Google Calendar for searching across platforms using plain language
Quick export sends Copilot responses directly into Word, Excel, or PowerPoint without copy-pasting
Windows Settings integration means you can adjust system preferences by describing what you want in your own words
Manus AI agent builds functional websites from files in a folder through a simple right-click in File Explorer
Gaming Copilot launches on ROG Xbox Ally handheld devices, offering tips without leaving your game
Privacy controls keep everything opt-in, with visibility into what the AI is doing and options to stop or disable features anytime

Voice interaction on computers isn't new. People have been using dictation, voice typing, and voice commands for years, yet keyboards remain the default for most tasks. Microsoft's observation that voice doubles Copilot engagement is interesting, though engagement by itself doesn't reveal whether people are accomplishing more or simply using the feature more often. The difference matters when considering whether these tools genuinely improve productivity or just change how we interact with our devices. Real-world use over the coming months will show whether voice becomes a preferred method or remains situational.

Copilot Vision's ability to see your screen eliminates the need to describe problems in words, which could be particularly helpful when learning unfamiliar software or troubleshooting technical issues. Instead of typing "how do I adjust the brightness in this photo editor," the AI can look at your screen and guide you directly. This convenience comes with the tradeoff of granting screen access to the AI, which some users may find uncomfortable regardless of Microsoft's privacy assurances. The feature being opt-in addresses this concern to some degree, but it places responsibility on users to understand what they're enabling.

The introduction of Copilot Actions and tools like Manus moves beyond conversational AI into territory where software takes autonomous action. Microsoft is being careful here, starting with preview releases and limited use cases while acknowledging the AI will make errors. This measured approach makes sense given what's at stake—these aren't just suggestions or answers, but actual operations on your files and system. Whether people will trust AI agents to work independently, even with supervision options, depends on how reliably these tools perform and how clearly they communicate what they're doing. The preview period will be crucial for understanding both the capabilities and limitations of this approach.

Read the full article here.

OTHER INTERESTING AI HIGHLIGHTS:

Spotify Partners With Record Labels To Create Artist-First AI Tools

/Spotify Newsroom

Spotify is joining forces with Sony Music Group, Universal Music Group, Warner Music Group, Merlin, and Believe to co-develop responsible AI tools that empower artists rather than compete with them. The collaboration focuses on transparency, consent, and fair compensation—ensuring AI deepens connections between artists and fans. The partners will co-create new products guided by four key principles: partnership, choice, fair pay, and creativity. Together, they aim to shape an AI future that protects artistry while unlocking new creative opportunities.

Read more here.

Claude Becomes More Customizable With Skills

/Anthropic Newsroom

Anthropic has introduced Claude Skills, a new way for users and developers to give Claude specialized abilities for specific tasks. Skills are modular folders containing instructions, scripts, and resources that Claude can load when relevant, improving performance for tasks like data analysis, brand writing, and file creation. The update makes Claude more adaptable and efficient across Claude apps, Claude Code, and the API—allowing teams to build, share, and deploy expertise with ease. Anthropic says this marks a step toward more customizable and collaborative AI workflows.

Read more here.

SOME AI TOOLS TO TRY OUT:

Alani Connect – Turns files and links into shared chat rooms for source-backed Q&A.
X-Pilot – Turns your ideas into professional educational videos in minutes.
Jungle AI – Generate flashcards and quizzes instantly to learn faster with AI.

That’s a wrap on today’s Almost Daily craziness.

Catch us almost every day—almost! 😉

EXCITING NEWS:

The Another Crazy Day in AI newsletter is on LinkedIn!!!

Another Crazy Day in AI: The Creative Control Dilemma in Generated Content

Oct 16
4 min read

Hello, AI Enthusiasts.

Hope your week’s been steady so far, or at least moving at a manageable speed.

Google DeepMind is cranking things up with the latest upgrade to its generative video model, giving creators more hands-on control over every frame, sound, and scene.

Meanwhile, Apple’s latest chip flexes serious power for AI running directly on your device.

And Anthropic’s lightweight model wants to make that power available to anyone, anywhere, in real time.

Just when you think it’s slowing down… it speeds right back up.

Here's another crazy day in AI:

Google launches Veo 3.1 with enhanced controls
Apple introduces M5 and pushes AI power further
Anthropic rolls out Claude Haiku 4.5 with major upgrades
Some AI tools to try out

TODAY'S FEATURED ITEM: Veo 3.1 Adds Audio and Editing Tools to Flow

Image Credit: Wowza (created with Ideogram)

How much creative control should you have over an AI-generated video?

Google DeepMind has introduced Veo 3.1, the newest version of its generative video model, now integrated with Flow, its AI-powered filmmaking platform. In a recent article on The Keyword, Jess Gallegos (Senior Product Manager, Google DeepMind) and Thomas Iljic (Director of Product Management, Google Labs) share how this update expands what creators can do within Flow — from controlling visuals and structure to integrating sound. Since Flow’s launch five months ago, it’s already powered over 275 million videos, and this latest release builds on that momentum with a stronger focus on precision and creative flexibility.

What's new in this release:

Sound enters the picture — Generated audio now enhances tools like “Ingredients to Video,” “Frames to Video,” and “Extend,” helping bring scenes to life with sound design that matches the visuals.
More control over visual details — Users can now combine multiple reference images to guide lighting, texture, and style, allowing for a more deliberate artistic direction.
Smoother storytelling through transitions — “Frames to Video” connects two stills into one continuous clip, useful for creating narrative or stylistic flow.
Extended shot creation — The “Extend” tool generates videos that can last a minute or more, maintaining continuity from one clip to the next.
In-platform editing capabilities — The new “Insert” feature lets users add new elements seamlessly into scenes, while an upcoming “Remove” tool will allow for natural object deletion without breaking the composition.
Access across platforms — Veo 3.1 is available through the Gemini API, Vertex AI, and the Gemini app, making its tools accessible to a wide range of users, from developers to enterprise teams.

The editing capabilities Google has added here speak to something anyone who creates content understands—your first version is almost never your final version. Being able to tweak and adjust generated clips without starting from scratch sounds useful on paper, but there's a gap between what's promised and what actually works when you're trying to finish a project. If you insert a character into an existing scene, will the lighting look right when you examine it closely? Can you remove an unwanted object from a busy background without creating visual artifacts that give away the edit? These are the kinds of details that matter when you're evaluating whether a tool fits into your workflow or just creates more problems to solve.

Then there's the audio situation. Anyone who's worked with video knows that silent footage only gets you so far. Having audio generate automatically means one less thing to handle separately, which is convenient. But Google calling these audio features "experimental and actively improving" is probably something to pay attention to. It hints that the results might be inconsistent depending on what you're making. Maybe it handles ambient sounds well but struggles with more complex audio. Maybe it works fine for rough drafts but doesn't quite reach the level you'd want for finished work. The honest answer is that it probably depends—on your project, your standards, and what you're willing to accept. Some people will find the generated audio perfectly adequate, while others will still end up doing custom sound work. That's not a criticism, just a reality of tools that are still evolving and trying to serve different needs at once.

Read the full article here.

OTHER INTERESTING AI HIGHLIGHTS:

Apple Introduces M5 and Pushes AI Power Further

/Apple Newsroom, Press Release

Apple has unveiled M5, its most powerful chip yet—built on third-generation 3-nanometer technology and designed to deliver the next big leap in AI performance. The new 10-core GPU includes a Neural Accelerator in each core, offering over 4x the peak GPU compute performance for AI compared to M4. With a faster Neural Engine, higher unified memory bandwidth, and the world’s fastest performance core, M5 brings significant boosts across MacBook Pro, iPad Pro, and Apple Vision Pro. The chip is purpose-built for on-device AI, enabling faster creative workflows, richer graphics, and more responsive Apple Intelligence features.

Read more here.

Anthropic Rolls Out Claude Haiku 4.5 with Major Upgrades

/Anthropic Newsroom, Product Release

Anthropic has released Claude Haiku 4.5, the newest version of its lightweight AI model that delivers near-frontier coding performance at a fraction of the cost. It runs twice as fast and one-third as expensive as Claude Sonnet 4 while even outperforming it in certain tasks like computer use. Haiku 4.5 is also Anthropic’s safest model to date, showing improved alignment and lower rates of concerning behavior in safety evaluations. Available now through the Claude API, Claude Code, and major cloud platforms, Haiku 4.5 makes high-performance AI more accessible for real-time and developer use.

Read more here.

SOME AI TOOLS TO TRY OUT:

Flint – Automatically builds on-brand pages tailored to your visitors and competitors.
Amplify – Repurpose long-form content into viral, brand-authentic LinkedIn posts.
Poke – Texts you when emails need replies or meetings are coming up.

That’s a wrap on today’s Almost Daily craziness.

Catch us almost every day—almost! 😉

EXCITING NEWS:

The Another Crazy Day in AI newsletter is on LinkedIn!!!