|
Hello Reader,
Promptcraft is a weekly curated newsletter on AI for education designed to elevate your AI literacy.
In this issue, you’ll discover:
- European Union Passes Sweeping AI Act, Sets ‘Global Standard’
- Florida teens arrested for creating ‘deepfake’ AI nude images of classmates
- A guide to Google Gemini and Claude 3.0, compared to ChatGPT
Let’s get started!
~ Tom Barrett
EU AI ACT
.: European Union Passes Sweeping AI Act, Sets ‘Global Standard’
Summary ➜ The European Parliament has passed the Artificial Intelligence Act, which will take effect later in the year. This landmark law is the world’s most comprehensive AI regulation and aims to protect fundamental rights, democracy, the rule of law, and environmental sustainability from high-risk AI while also encouraging innovation. The law bans certain AI applications that threaten citizen rights and establishes transparency requirements for general-purpose AI systems. The law aims to make the EU the de facto global standard for trustworthy AI, and lawmakers say it is only the first step in building new governance around technology.
Why this matters for education ➜ When we take a global view of the changing nature of technology regulation, all educators are impacted by this new law. The second-order effect in other parts of the world might include similar, tighter regulations on high-risk AI applications. As the most comprehensive AI regulation to date, it sets a global standard for developing and deploying AI technologies, including those used in educational settings. This new law will likely influence the direction of AI innovation and regulation worldwide as governments and organisations seek to establish their own guidelines and recommendations. Although it might be an oversimplified way to consider the impact, one question we might ask is how innovation in the educational technology space will be encouraged or stifled.
|
DEEPFAKE
.: Florida teens arrested for creating ‘deepfake’ AI nude images of classmates
Summary ➜ Two middle school students in Florida have been arrested and charged with third-degree felonies for allegedly creating deepfake nude images of their classmates using an unnamed AI application. This marks the first instance in the US of criminal charges related to AI-generated nude images. The incident highlights the increasing problem of minors creating explicit images of other children using generative AI, with only a handful of states having laws addressing this issue.
Why this matters for education ➜ Similar to the story from Los Angeles a little while ago, there is no need to explain why. While President Joe Biden has issued an executive order on AI banning the use of generative AI to produce child sexual abuse material, there is currently no federal law addressing nonconsensual deepfake nudes. As a class, school or system leadership team, you might pause and consider how you would respond if this scenario played out in your community. What policies and procedures should we implement to ensure we are prepared to handle instances of AI technology misuse within our school community?How can we foster an open and supportive culture in which students feel comfortable reporting such issues, and what support systems can we establish to assist students who may become victims of these actions?
|
|
FRONTIERS
.: Your guide to Google Gemini and Claude 3.0, compared to ChatGPT
Summary ➜ Two new powerful language models, Google’s Gemini Ultra 1.0 and Anthropic’s Claude 3.0 Opus, have been released, rivalling OpenAI’s GPT-4. This article compares the models and provides strategies for organisations deciding which to use, ranging from using just ChatGPT to adopting all three. The release of these models is a milestone, giving developers more choices, affecting company revenues, and indicating the difficulty of surpassing GPT-4 level performance.
Why this matters for education ➜ This article compares frontier AI models and provides helpful ideas for educators looking to improve their AI literacy. Hands-on experience with leading proprietary models like Gemini, Claude, and GPT is critical to understand their capabilities and potential applications in the classroom. System-wide decisions about AI tool rollouts in schools may depend on existing technology ecosystems, with schools potentially leaning towards tools that integrate seamlessly with their current setup. However, understanding the strengths and weaknesses of each frontier model can help educators make informed decisions about AI adoption, regardless of existing partnerships.
Complement this with my take on open source models below in the reflection titled: Peanut Butter and Pickles: Can Open Source and Education Mix?
|
.: Other News In Brief
🔓 Should AI be open?
🎓 How Young Is Too Young to Teach Students About AI? Survey Reveals Differing Opinions
💕 Why people are falling in love with AI chatbots
🚫 Google Bans U.S. Election Questions in Gemini AI
👨💻 Cognition launches an AI software engineer agent, Devin
🫂 Empathy raises $47M for AI to help with the practical and emotional bereavement process
🛠️ Microsoft opens its Copilot GPT Builder to all Pro subscribers
🖼️ Midjourney debuts feature for generating consistent characters across multiple gen AI images
🏢 OpenAI CEO Altman wasn’t fired because of scary new tech, just internal politics
🆓 Elon Musk vows to make his ChatGPT competitor Grok open source
|
:. .:
Monthly Review
.: All the February issues in 1 PDF
 |
Promptcrafted February 2024
The only monthly publication that curates the most relevant and impactful AI developments specifically for educators.
A… Read more
|
|
.: :.
What’s on my mind?
.: Peanut Butter and Pickles: Can Open Source and Education Mix?
One area of educational technology that seems to be overlooked is the potential of open-source software and tools.
Open source means the source code is freely available to the public to view, modify, and distribute, encouraging collaborative development where anyone can contribute improvements or modifications to the project.
However, in my experience, open source has never been an option in educational technology strategies in schools and systems.
This raises the question: Is education missing out on the benefits of open source?
Back when we were still learning to use Word Processing software and set up computer labs in our schools, I remember coming across an open-source version of MS Word called OpenOffice.
It was a suite of office productivity tools that was an open-source alternative to Microsoft Office. OpenOffice was free to download and did almost everything the licensed Word version could do. But nobody knew about it, and open-source software was never seriously considered.
Perhaps education and open source don’t go together like mince in a trifle or peanut butter and pickles. I mean, I like all of those things, but not together.
While open-source tools provide the flexibility to customise and adapt to specific use cases, this freedom can lead to application inconsistency and a lack of standardisation. This can pose challenges in an educational setting, where uniformity in tool usage is often what system admins want to maintain.
Additionally, the open nature of these tools can sometimes pose security concerns, as the code is accessible to everyone, including potential malicious actors.
The benefits of open source cannot be ignored. Within the AI space, there are a vast number of open-source models that can be used for free.
At the time of writing, there are 548,994 models on Hugging Face for a wide range of multimodal, computer vision, natural language processing, and audio functions. Yet, we might only know about ChatGPT and Gemini for everyday users and educators.
So, the challenge is educating the education sector about these open-source models’ existence, benefits, and potential drawbacks. This involves raising awareness, providing clear and accessible information about implementing them and offering guidance on managing any possible risks associated with their use.
It also requires a shift in mindset from being reliant on big tech vendors to being open to exploring other options that could offer greater flexibility and adaptability.
Will there be resistance to open source? Does anyone know about it? Are we so wedded to big tech vendors we can’t see other options?
What do you think?
:. .:
~ Tom
|
Prompts
.: Refine your promptcraft
It is becoming clearer that effective prompcraft falls into three broad approaches.
- Start small and iterate.
- Structured longer prompts.
- Build an in depth system prompt for a custom bot.
I use all three of these in my daily interactions with various tools.
Let’s look at how to start small and iterate with the Flipped Interaction prompt.
This is when you ask the LLM to ask you questions before it provides an output. This helps build contextual cues and information.
According to research referenced by Briana Brownell from Descript:
for the highest quality answers, the tests showed the Flipped Interaction pattern is the valedictorian of prompts. […] In tests, using this principle improved the quality of all responses for every model size. It improved quality the most for the largest models like GPT-3.5 and GPT-4, but it did impressively well in smaller models too. So if you’re not using this A+ technique yet, you should definitely start.
Here are some example prompts to try:
Let’s collaborate on (describe your task) start by asking me some questions.
From now on, I would like you to ask me questions to (describe your task)
I would also recommend adding in an instruction to go one step at a time, or to limit the number of questions as most models are too verbose.
Here is an example of this using the new Claude-3-Opus-200k model.
And here is the same prompt using Alibaba’s Qwen-72b-Chat model.
:. .:
Remember to make this your own, try different language models and evaluate the completions.
Do you have a great prompt you would like me to share in a future Promptcraft issue? Drop me a message by replying to this email.
|
Learning
.: Boost your AI Literacy
DESIGN
.: Design Against AI: 2024 Design in Tech Report by John Maeda
John Maeda’s report advocates for designers to develop AI literacy while staying grounded in human reality, in order to help shape an ethical and human-centred future.
Balancing the use of AI as a tool with uniquely human creative abilities will be an ongoing challenge.
Reminds me of the premise of the humAIn community
A learning community for educators to connect and explore our humanity in the age of artificial intelligence
HUMANITY
.: AI Literacy is the Art of Synergizing Intuitions
“Dealing with AI is a wandering dance between two unconscious entities—the AI and much of our brain—using the much tinier piece of our neural mush that deliberates.”
This quote captures the central analogy of the post – that interacting with AI is an interplay between the intuitive, automatic responses of both human and artificial neural networks, mediated by our limited conscious faculties.
Tim Dasey’s article frames AI literacy as a process of syncing the intuitive, contextual responses of both human and artificial intelligences through techniques like prompting, feedback solicitation and comparative understanding of cognition.
Mastering this “dance” unlocks AI’s potential while honing essential human skills.
COLLECTION
.: The AI Literacy Curriculum Hub
The AI Literacy Curriculum Hub is a spreadsheet curated by AI for Equity and Erica Murphy at Hendy Avenue Consulting.
A collection of AI literacy lessons, projects, and activities from respected sources like Common Sense Education, Stanford’s Craft AI, Code.org, ISTE, MIT Media Lab, and more.
Each resource is tagged, providing key details such as the applicable grade levels, lesson duration, required materials, and learning objectives.
|
Ethics
.: Provocations for Balance
➜ Is it time for schools to become digital dictatorships, monitoring every keystroke and thought, or do we resign ourselves to a future where trust is a relic of the past?
➜ Big Tech AI companies lure schools with promises of personalised learning and cutting-edge tech, while open-source alternatives whisper seductively of freedom and transparency. In this high-stakes game of AI roulette, who will educators bet on? Will they sell their digital souls for a taste of Silicon Valley’s forbidden fruit or take a leap of faith into the wild west of open source, where danger and opportunity lurk in equal measure?
Inspired by some of the topics this week and dialled up.
|
:. .:
.: :.
The more we invest in our understanding of AI, the more powerful and effective our education ecosystem becomes. Thanks for being part of our growing community!
|
