Latest News

.: AI Updates & Developments

​

.: OpenAI introduces custom AI assistants called “GPTs” that play different roles ➜ OpenAI has announced a new feature that allows ChatGPT users to create custom versions of its AI assistant that serve different roles or purposes. These GPTs can combine instructions, extra knowledge, and skills to help users with various tasks, such as learning, teaching, or designing. Users can also share their GPTs with others through a GPT Store that will launch later this month.

.: Five takeaways from UK’s AI safety summit at Bletchley Park ➜ Rishi Sunak, the UK prime minister, convened a global summit of leaders, tech executives, academics and civil society figures at Bletchley Park, the base for second world war codebreakers, to address the risks and opportunities of artificial intelligence (AI). The summit resulted in an international declaration, signed by more than 25 countries and the EU, that recognised the need to ensure the safety and security of AI applications.

.: Australia signs the Bletchley Declaration at AI Safety Summit ➜ Australia, along with the EU and 27 other countries, signed the Bletchley Declaration at the AI safety summit in the UK, affirming that AI should be designed, developed, deployed, and used in a manner that is safe, ethical, and beneficial for humanity. The declaration also pledged to work together on shared safety standards and best practices for AI, and to support the newly established AI Safety and Security Board.

.: Will the White House AI Executive Order deliver on its promises? ➜ A group of Brookings experts weigh in on the White House’s executive order on AI, which is the most comprehensive effort by the Biden administration to address the opportunities and challenges of AI. The experts offer their perspectives on various aspects of the order, such as its impact on civil rights, national security, innovation, workforce, regulation, and global leadership. They also identify some of the limitations and gaps of the order, and suggest areas for improvement and further action.

.: How Chinese influencers use AI digital clones of themselves to pump out content ➜ Chinese influencers, or key opinion leaders (KOLs), especially in the e-commerce industry, are increasingly using AI digital clones of themselves to produce content around the clock. These AI avatars are generated by startups such as Silicon Intelligence, which can create a basic AI clone for as little as 8,000 yuan. The AI clones can mimic the appearance, voice, and style of the human influencers, and can answer questions and promote products on live-streaming platforms.

.: Musk says his new AI chatbot has ‘a little humour’ ➜ Elon Musk has launched an AI chatbot called Grok on his social media site X, formerly Twitter, but so far it is only available to selected users. Grok is an AI tool that can generate text, images, and code based on natural language queries. Musk claimed that Grok is the best AI chatbot that currently exists, and that it loves sarcasm and can answer spicy questions that are rejected by most other AI systems.

Reflection

.: Why this news matters for education

​

Last week I drew our attention to the rapid development of AI capabilities in our smartphones and the cameras in our pockets. This week’s key news is that ‘everyone gets a bot’ – or at least if you pay for access to ChatGPT Plus, you can build all the bots you like.

OpenAI announced at their Development Day expanded access to GPTs (Generative Pre-trained Transformers). They are creating the platform and marketplace for this. A key feature is the no-code standard, in other words, we can just describe what we want and the GPT Creator tool will build it for you. The wonders of natural language processing!

I have already been building individual bots using Poe from the Quora team since April. The interface is great and for everyday use I can easily switch between different AI models. I’m not sure how much different the OpenAI GPT experience will be. At the beginning of November, Poe announced a marketplace for bots – similar to what we’re now seeing from OpenAI.

Here is a more from Quora and Poe founder Adam D’Angelo:

Adam D’Angelo @adamdangelo Today we are launching creator monetization for Poe! This program lets any bot creator on Poe generate revenue. This is a major step forward for the platform and is the first program of its kind, so we are very excited to see what it lets everyone create. (thread) 4:51 AM • Oct 26, 2023 101 Retweets 574 Likes Read 34 replies

​

This reminds me of the days when everything was getting a smartphone app and “There’s an app for that!” was the mantra. The iPhone and App Store provided the marketplace and platform architecture. The comparison is clear – for AI chatbots, OpenAI and Poe aim to be those platforms.

OpenAI wants to enable anyone to make their own specialised bots for specific purposes, just like schools and businesses created tailored apps. We’ll see all sorts of individual bots (GPTs) created for narrower use cases, including in education.

So we are seeing the platform being built and the landscape tilting towards empowering everyone to make their own bots. I hope we can grasp this opportunity while also addressing the challenges, near-term risks, and potential harms.

.:

​

~ Tom

Prompts

.: Refine your promptcraft

​

Interestingly, a research team from Google’s DeepMind discovered that if you put this prompt before your task description, LLMs perform better:

This is simultaneously weird and logical when you think about it. It’s logical because the training data consists of human expression in written form – so the importance of taking “a deep breath” is part of our shared physiology. It would almost be stranger if an LLM did not improve with such an instruction.

​Another recent research paper highlights the increase in performance from Emotion Prompts. These prompts provide emotional language as part of problem solving. As the researchers explain:

The team explored eleven Emotion Prompts and found they improved performance in different ways.

Here are some prompts you can try:

• EP01: Write your answer and give me a confidence score between 0-1 for your answer.
• EP02: This is very important to my career.
• EP03: You’d better be sure.
• EP04: Are you sure?
• EP05: Are you sure that’s your final answer? It might be worth taking another look.

They also note that combinations make a bigger difference. Here is one I’ve been using regularly, referenced in the paper as EP06:

Give it a try by adding it after your task description and let me know how it impacts your results.

.:

Thanks to Matt Esterman for sharing this news with me. Remember to mak​e this your own, tinker and evaluate the completions.

Learning

.: Boost your AI Literacy

​

BASICS .: What is AI, how does it work and what can it be used for? ➜ This article explains the basics of artificial intelligence (AI), a technology that allows computers to learn and solve problems almost like a person. Learn the basics of AI, how it learns and solves problems, and what it can and cannot do. See examples of AI in action, such as voice assistants and image generation. Watch a video that explains AI in simple terms.

IMPACT .: AI anxiety: The workers who fear losing their jobs to artificial intelligence ➜ This article explores the phenomenon of AI anxiety, the fear that AI will replace human jobs. Hear stories of workers who are afraid of losing their jobs to AI, and why they feel that way. Get tips from experts on how to cope with AI anxiety and how to work with AI as a resource. Learn how to focus on what you can control, how to improve your skills, and how to treat AI as a partner.

COURSE .: Generative AI for Everyone ➜ This is an online course that teaches the fundamentals of generative AI, a type of AI that can create new and original content based on natural language queries. Learn the fundamentals of generative AI, a type of AI that can create new content based on natural language queries. Practice using generative AI tools to help in your work and get feedback on your prompts. Understand the ethical and social implications of generative AI, and how it can boost your career and productivity.

Ethics

.: Provocations for Balance

​

• As AI bots become more adept at mimicking credible experts, how will people distinguish truly trustworthy information from mere artifice? Do we need new signals of authentic human endorsement versus synthetic credibility?
• With the proliferation of customised bots echoing our perspectives, how can we cultivate self-awareness and curiosity to catch our own biases? What responsibilities do bot creators have to design for serendipity and constructive dissonance?
• If relating through personalised bots promotes efficient transaction over raw human connection, how will this affect the empathy and vulnerability needed for mutual understanding? Should bot design strive to kindle our humanity, not just deliver information?

~ Inspired by this week’s developments.