History Teacher looking to the past | Developer looking to the future. ✍️https://xitter.cf/t.co/QZYH0QH7wZ 👶https://xitter.cf/t.co/3Cj6Om1QML 🍽️https://xitter.cf/t.co/xLvFmE5p0s 📝https://xitter.cf/t.co/YenA9vPY1Z
England, United Kingdom
Joined October 2011
- Tweets363
- Following996
- Followers99
- Likes2.3K
Pinned Tweet
Summer holidays = coding season.
Last term I have really enjoyed building a feedback and marking app to try and help save teachers hours of work. Now I’m finally cracking on with the other ideas sitting in my notes.
School starts again, and we’re expecting a baby, in September...
Better get building now while I have the chance...!
Developer Dan retweeted
📘 My new Teaching for Attention Playbook is a 70+ page, evidence-informed guide that breaks down the science of attention and provides 6 practical strategies to support teachers.
It covers how to:
🎯 Secure Attention: Create the conditions for students to focus
🎯 Direct Attention: Guide students towards important information
🎯 Orchestrate Attention: Manage where attention goes as learning unfolds
👉 Want a free copy? REPOST and comment ATTENTION below, and I’ll DM you the link.
⏰ Available free until Wednesday 30 September at 11:59pm AWST
What's the best book you have read on AI?
I've got a bit more time over the next few weeks before work starts again and I'm wanting to do a bit mroe reading before this. I've started a few books recently but even ones published 6 months ago feel outdated with the current pace of AI at the moment.
Has anyone read anything really good that stays relevant with the newest releases?
Finally into developing for the latest iOS and must say the new Siri app is aesthetically beautiful. Eco sites to try out and see if it is more usable now
Ok, Codex is telling me this will take six to 9 engineering days.
It's a big implementation I am planning so how long do we think it will take Sol Ultra on fast mode?
Lets see...
This will give companies the dilemma of not having their data used effectively to train models Vs being able to accurately show up in LLM responses…
This font looks perfectly normal to humans but wreaks havoc on AI fastcompany.com/91585494/thi…
Here’s hoping the review process is quick as the last one to fix some small bugs before users download the previous one!
As AI development accelerates, states and organisations may treat staying ahead as a security necessity which will turn technological competition into a form of cyberwarfare. Therefore, that same intelligence that can discover a vulnerability can also build an exploit or produce a patch, so advantage may belong less to one permanent side than to whoever can organise and act first.
Advanced capabilities require strong safeguards. That’s why access is limited to approved defenders, with additional controls and monitoring for higher-risk cybersecurity work.
openai.com/index/expanding-d…
Finally getting to a point where I am happy with the little things as well as the big things with the UI and UX
Although I'm already down to 65% with some conservative use since the reset yesterday and hoping for @thsottiaux's promised performative reset today...
One of the biggest benefits of using 5-10 threads at once on Codex is it actually means there is less need for fast mode and so it saves my usage a bit too.
Very interesting to see their attempts at safety here given recent news. I am loving the consistency in naming models with the space theme and keeping to this with Day break. Going to give these an ago and see what kind of vulnerabilities it can spot in my existing software.
I think he take of ‘browser : AI :: phone : smartphones’ if very on the mark. While vibe coding has exploded, a lot of the flood of apps are basically LLM wrappers. I don’t actually have a problem with this as in the most basic form, most SaaS is just one form of software wrapped in another. But I do think the success of LLM wrappers show that people like to interest in it in different ways. That’s why I think the first company to perfect the fully user interface will jump ahead. The way we ‘browse’ LLM content will be key
ChatGPT’s Atlas browser was sunset today. Perplexity’s Comet browser is in zombie mode.
When we started @browsercompany nobody cared about browsers, then everyone did (copying Dia), then everyone bailed.
Yet I’ve never felt more confident browser : AI :: phone : smartphones.
I think this make sense for two reasons. One is the basic change in people using LLMS over search (not everyone but enough to amke a difference) and the second is that when using an LLM over search, it can easily search 10, 20, 50 sites in the time it would take a human to search and read one so it naturally increases traffic further. This is just one example.
been busy implementing some beautiful frames and edges to the components in this app thanks to inspiration from @Jakubantalik. If you haven't checked out transitions.dev then go for it - some really helpful ideas!
its been a while but next updates submitted! Hopefully the @Apple review process will be quick
What percentage of X posts are made purely because people are waiting around for their agents to finish tasks?
can't believe I am only jsut testing out the benefits of @RevenueCat now. I have been Apple piled for so long that I assumed the beasiest and best way was through all their own APIs. One day working with RevenueCat and I can see hwo wrong I have been