💡Bonus Tip: Zombie App Audit - (apps don't use but with location, contacts, and camera + microphone permissions). The Hack: Go to phone settings: "Unused Apps" and Offload them. Save storage space but keep your data - if you need them again they're a click away. #DigiBlogChat

A10. My goal is to spend less time "scrolling" and more time "creating." Part of that is setting stricter App Limits and making sure my digital space reflects my actual priorities. #DigiBlogChat

A9. I like the 3-2-1 Rule: 3 copies of your data, on 2 different media types, with 1 kept off-site (cloud). If your photos only exist on one hard drive, you’re one spilled coffee away from a heartbreak. Doing it is more challenging! ☕️💻 #DigiBlogChat

A8. I’m increasingly using Notebook LM. Having a "Second Brain" to dump ideas into means I don't have to keep 50 tabs open or 100 sticky notes on my physical desk. I can generate slides, podcasts or explore ideas. #DigiBlogChat

A7. I’m a fan of the "Friday Reset" (any day could work) - every Friday I empty the trash, purge emails, review data stores and ruthlessly delete unused or out of date content. It makes Monday morning feel like a fresh start instead of a pile of chores! #DigiBlogChat

A6. Absolutely. If an account no longer inspires me or just adds noise to the feed, I unfollow or mute. My feed is my digital environment; I want it to be a place I actually enjoy spending time. #DigiBlogChat

A5. Spring cleaning is the perfect time to audit your Password Manager. I look for any "reused" password alerts and update them. Also, check which third-party apps still have access to your Google or X accounts—you’d be surprised! #DigiBlogChat

A4. Be ruthless! You can always find most content again, why store it? For the keepers, I use a simple YYYY-MM-Event naming convention for folders so they’re actually searchable. #DigiBlogChat

A3. I’m a big fan of Unsubscribe. If I’m just hitting "mark as read" every day, it’s time to part ways. It’s not "ghosting" if it’s a brand! #DigiBlogChat

A2. I always start with the Downloads folder. It’s the "junk drawer" of the internet. If I haven’t looked at that PDF invoice or random meme in three months, it’s going straight to the bin. #DigiBlogChat

A1. To me, it’s all about digital hygiene. It’s that once-a-year (or once-a-quarter) deep dive into my files, apps, and subscriptions to delete the "clutter" that's slowing down both my devices and my brain. Clear drive, clear mind! #DigiBlogChat

Between questions, we’d like to announce next week’s topic: Build relationships with your most engaged followers! With Samir of @fedica.com ! #DigiBlogChat

Join us Tuesday, March 10 for Digital Spring Cleaning Hacks! #DigiBlogChat Times: 7p UK, 8p W Europe, 3p ET, Noon PDT! Note: Most in N America change their clocks 3/8. For those in Europe the chat is 1 HOUR EARLIER for the next 3 WEEKS. Questions: yoursocialmediaworks.com/digi-spring-...

🚨 Check out the questions 👇 which we will discuss in some minutes .. ready, steady, go ..! 📱 @carolstephen.bsky.social and @lazblazter.bsky.social are welcoming all of us in less than 10 min. to the amazing #DigiBlogChat 🙌 See ya ..!

I hear „decluttering“ and I am all in! 🧹📧 Let‘s talk e.g. about our unread emails in the inbox caused by newsletters • any other suggestions? 💻 📱 📵 Who joins in today • tonight? Such a great #DigiBlogChat • Meet you there! 😆

Sincere thansk to @carolstephen.bsky.social for today's question set and inspiring us all to share ideas. Best wishes to all and thank you for supporting #DigiBlogChat

A10. I don't maintain a journal personally, though I can see that Social Media offers a modern way of sharing personal events against a timeline. I'll offet that a journal articulates experiences from your own perspective and "files" them for future reference, a bit like photos. #DigiBlogChat

A9. I've re-read the Iliad and The Odyssey many times. What always occurs to me is the detail applied to each of the principal characters and how it has inspired so many to seek definitive proof of this chapter in history. Have any other books had such an effect? #DigiBlogChat

A8. Conversations with the dead are something I would have shied away from - too many horror film influences. Losing Mum gave me a different perspective though and I keep her in my thoughts with regular family updates. #DigiBlogChat

A7. I read a lot via Kindle these days so will take a peek at recommended books at the end of a read. I also note when new trusted author publications are due out and may pre-purchase. #DigiBlogChat

A6. The "24-Hour Rule." Within 24 hours of reading a great idea, you must do one small thing to act on it. #DigiBlogChat

A5. Active Recall. Close the book and write down: "What was the most important thing I just learned, and why does it matter to me?" #DigiBlogChat

A4. When it comes to intent and genre. Publishers spend a lot of money making sure the "vibe" of the cover matches the content. I usually select by known Authors though. #DigiBlogChat

A3. This sums me up: Fiction: Read for immersion and empathy. Nonfiction: Read for utility and structure. Skim the Contents then dive deep into the chapters that offer immediate value. #DigiBlogChat

A2. Re-reading a favorite is like visiting an old friend—you know the way home, so you can focus on the scenery you missed the first time. I think my most re-read books are the Iliad and The Odyssey. #DigiBlogChat

A1. I don't tend to annotate books as such. I will use post it notes in drafts for work and using apps like One Note has been useful. Now my go to is Google Notebook LM though, simply a great way to store and compare information. #DigiBlogChat

Hi @carolstephen.bsky.social and #DigiBlogChat - so sorry I can't be live today but been whisked off to watch Diversity at the Bath Forum this evening. I have prepared my answers though and look forward to reviewing the questions and responses tomorrow. Have fun.

You may find this interesting if you've not heard of AI temperature. Higher the setting the higher the creativity and likelihood of hallucination. Microsoft set their Copilot at 70% (balance and reliability) whereas OpenAI the full 100% (1.0). Hardly ever mentioned but explains a lot.

Thanks to all for today's #DigIBlogChat - hopefully the question set was stimulating and not distressing. My hunch is that the types of issue debated today are necessary to ensure that we stay focused on tech for good ultimately. Take care everyone, see you on a post somewhere. #DigiBlogChat

A big thank you to @lazblazter.bsky.social for the topic and questions! We've all had our little gray cells stretched today! #DigiBlogChat

Always food for thought in our weekly chats, let me know if we need something for next week too. Thanks to all our participants and to you for scheduling and guiding the chat. #DigiBlogChat

Yes, that sounds right, John! Sounds like a terrifying idea, though. #DigiBlogChat

A9 Deleting an entire model may be disproportionate. Courts could require “machine unlearning,” dataset removal, damages or licensing instead. The remedy should fit the harm. AI law will likely evolve toward surgical fixes, not digital death penalties. #DigiBlogChat

A10a A synthetic influencer crosses the line when it causes consumer confusion or exploits likeness rights. In cases involving publicity rights, courts often ask: Would a reasonable person think this is endorsed? Parody has room. Deception does not. #DigiBlogChat

A10b Intellectual property in the AI era is less about fences and more about governance. Code, prompts, style, likeness, data — all blur. The winners won’t rely on one doctrine. They’ll combine law, contracts, tech controls and culture to protect what matters. #DigiBlogChat

And depending on where you are, the whole point of intellectual property protection is probably moat. #DigiBlogChat

Maybe, Carol. But if learning is generally about forming associations between things, those can become engrained in ways which are hard to undo selectively. #DigiBlogChat

A6. AI companies in America have been pretty successful in claiming that reading material is fair use. The fairness seems very one sided though whereas it's definitely capable of causing damages through recognising and repeating patterns. #DigiBlogChat

A8. Interesting question as there is value in the output by a machine which does not have human protections and whereas the prompt is human generated. Probably needs accepted case law. #DigiBlogChat

A8 A complex prompt can qualify as a trade secret if it provides economic value and stays confidential. Once shared publicly, protection fades. Think of it like a secret recipe: value comes from controlled access, not the idea alone. #DigiBlogChat

A9 #DigiBlogChat I'd bet that, as with humans, AI models find it harder to unlearn than to learn.

A10. Understandably emotive. It crosses the line when it achieves a level of commercial substitutability in place of a real person and their attributes. Key Takeaway: Parody is protected; replacement is a lawsuit waiting to happen. #DigiBlogChat

A9. Technically it's hard to "delete" a specific memory from an LLM without breaking the rest of it. Total deletion is the "nuclear option" that courts will avoid unless the infringement is systemic. Don’t burn the library because one book was stolen; try to recover the stolen pages. #DigiBlogChat

Q9. The Right to Unlearn: If a court finds that an AI was trained on infringing material, is the only remedy to delete the model entirely? #DigiBlogChat

Q10. Digital Twins vs. Fair Use: At what point does an AI-generated "synthetic influencer" that mimics a celebrity's cadence and personality cross the line from parody/homage into IP theft? #DigiBlogChat

A8++ Interesting, Jim. Recently I asked a guy who'd patented an AI-based process about whether it wouldn't be better to keep it secret ... unless you can police and defend it. ;-/ #DigiBlogChat

A7. There are deep learning models that have successfully used unstrippable watermarks, such as “DeepSigns,” which inserts watermarks within layers of a .pdf. #DigiBlogChat

A8. If a prompt is complex enough to produce a consistent, unique result, it functions like a chemical formula. If someone "reverses" your prompt through trial and error, proving they stole your specific string feels impossible. Complex prompts are the new "Source Code." #DigiBlogChat

Q7. Watermarking Vulnerabilities: Since AI can now strip invisible watermarks from images and code, what "un-strippable" markers can be developed to prove ownership deep within a model’s weights? #DigiBlogChat

Q8. Prompt as Trade Secret: If a company develops a highly complex multi-token prompt that allows an LLM to generate perfect proprietary code, is that prompt itself protectable IP? #DigiBlogChat