🤝 AI deal: US-China summit could ban autonomous weaponry

Captain’s Log… AI is moving at the speed of thought! 

Meta has unveiled an AI that can read human thoughts and translate them into images. OpenAI set the bar incredibly high with GPT4 Turbo and capabilities that effectively leapfrog many other AI startups — they’ll now need to pivot or perish. Plus, we’re seeing an accelerating convergence of AI and robots.

Even the former generation of tech pioneers are coming out of retirement. The founders of MySpace (a precursor to Facebook, for our younger readers) have a new AI company, Plai. Their first product, PlaiDay, is a text-to-image and video service that takes users’ selfies and creates decent avatars. I appreciate the name Plai, but feel the name “MaiSpace” would have been even better, lol.

For those not familiar with me, my name is Cory Lopes-Warfield – I’m a serial tech founder, EIR, investor, educator and influencer in the AI/blockchain/robotics space, and my primary platform has been LinkedIn, where I’m a power user; but I also have created some of the other biggest influencers on LinkedIn as a coach. 

I mention that only to say that OpenAI has now given users the ability to create and train their own GPTs, along with an App Store where creators will be able to monetize their GPTs. So I created an “Influencer Coach” GPT that’s pretty good, and its getting better all the time. 

Today, I’ll teach you how to create your own. 

Last week, I also taught you how to “write a book” using AI/ChatGPT, and “ate my own dog food” to write “The Hitchhiker’s Guide to Linkedin,” now available as a free e-book. 

That’s all to say, the tech exists and works, and people like me don’t just talk about it, we use it every day. And to the end, I proudly present AI Logs Issue 5. As always, we value and welcome your feedback.

NEWS

• 👎 Does AI help solve problems? Scientists at Hong Kong Laureate Forum say no The “Science for Diplomacy” initiative is another way the Forum is advancing social impact.

• 💠 Autonomous lab helps find quantum dots with highest yield in hours It takes humans years to achieve the same goal, but autonomous labs can work much faster.

• 🧐 Research finds gullible humans to be often tricked by AI faces Those beautiful faces you see in pictures online? They probably aren't real. Especially if they are white.

• How is ChatGPT biased? Researchers identify a variety of concerns Studies confirm that ChatGPT, the world’s most popular AI language model, has a number of biases in its responses.

• 💚 New AI tool can predict the risk of deadly heart attacks 10 years out The AI tool can also provide information on narrowing of the arteries and other clinical risk factors.

🛑 US and China set to forbid use of AI technology in autonomous weaponry

Presidents Biden and Xi are expected to announce a historic pact banning autonomous AI-powered weapons at today’s summit in San Francisco. As superpower tensions simmer, the leaders appear poised to pledge against applying AI for nuclear command/control and autonomous drones or robots. While details remain unconfirmed, the move would help restrict military AI applications to reduce catastrophe risks.

Both nations have raised concerns over unchecked AI development. Earlier this year, the US led an international declaration promoting responsible AI use in defense contexts. For its part, China has heavily invested in AI research, including for its military. However, leaders seem to recognize the dangers of delegating life-or-death decisions to AI systems. An AI weapons ban would be an important symbolic step toward cooperation.

A ban alone cannot fully prevent military AI abuses or arms races. But combined with ethical norms and transparency measures, it provides a foundation. As AI capabilities accelerate, more complex challenges will emerge around verification and international compliance. Yet the pact would establish key principles and continue important US-China dialogues. 

However, 80 percent of respondents in our Monday Daily Blueprint poll expected the development of autonomous weaponry to continue in secret, regardless of any agreements.

Click below to get the full story.

Other Important Updates

• New ‘AI chemist’ could autonomously make oxygen from ice on Mars It could enable autonomous oxygen production from Mars ice without human intervention.

• Samsung jumps on the wagon, unveils ChatGPT-like AI ‘Gauss’ The generative AI model can code, enhance images, and write emails.

• Amazon to unleash ‘Olympus’ against ChatGPT Amazon is reportedly investing millions in training Olympus.

• NVIDIA launches latest supercomputer AI chip H200 is the first GPU to offer the faster and larger HBM3e memory to fuel the acceleration of generative AI.

Prompt of The Week

Criticize Mode: Once you’ve got results such as email titles, book title suggestions, social media content, startup ideas, etc., from your AI, ask it to, "Act as a critic. Criticize these (results) and tell me why they’re bad.” 

You can also ask it to act as a harsh critic, and provide brutally honest feedback. This is a powerful prompt, but please be aware that it will continue criticizing itself until you tell it to stop doing so.  

AI Image of the Week

Last week, we covered the launch of xAI’s Grok chatbot. X user @KettlebellLife asked Grok to describe itself and “it went pretty hard,” resulting in this menacing self portrait produced in DALL•E3. The post has since collected nearly 679,000 views and 2,600 likes.

TUTORIAL

Create your own GPT (requires GPT-Pro subscription):

Create your own GPT (requires GPT-Pro subscription)

1. Go to “Explore” in the search bar, above your list of recent threads

2. Click “Create a GPT (beta)”

3. You’ll now be in “Create” mode, where you can chat with ChatGPT about what you want your GPT to do, what you will train it on, how it should behave, what users will be using it for, etc. 

4. You will then start engaging with it as though it were your own Chatbot to train it. You can upload files, videos, graphics and your own KB (knowledge base). It will also generate an icon button for you, and you can change it as desired.

5. In “Configure,” you can name it, amend the instructions it will be auto-generating, toggle features like Browser Mode, and add or amend the conversation starters it automatically generated for you.

6. Preview is where you can see the bot as others will see it, and play with it.

7. Save your GPT and make it private, accessible by link, or public for the world to enjoy!

🤑 You will also be able to monetize your best GPTs soon.

TOOLS OF THE WEEK

🤖 CraftJarvis/ Jarvis-1 is an open source project currently on Github that learns as it goes (similarly to MemGPT I’ve featured here before). It can also deploy simultaneous autonomous agents, it’s multi-modal, and it’s an all round huge step for AI (other than the less-than-original name – what’s with all the bad names in AI?). It has already figured out and is able to play Minecraft on its own already, which I find to be super cool!  The end goal is the execution of complex tasks on the user’s behalf, with minimal or no input from the user, using its own “memory.” Just to be clear, this Jarvis is not associated with either Microsoft’s or the Iron Man OS.

⏩ ChatGPT-4 Turbo: I have no option but to feature this, though you’re all likely to be at least somewhat familiar with it. This latest update legitimately takes ChatGPT to the next level. Here’s what’s new:  You no longer needing to toggle between DALL•E, the web browser, Data Analytics, or Vision, which alone is HUGE.  It’s faster, more accurate and can handle significantly larger character input.  It’s trained on much more recent data than previous versions. GPT creation (as described above) and an App Store for them.  It allows collaboration, JSON outputs, and even more.  This is close to what many have speculated a GPT-4.5 would be like, and it’s incredible.

📍 AI Pin from Humane: This will be released publicly tomorrow and is former Apple designers’ and engineers’ attempt at making the smartphone obsolete. Powered by T-Mobile, this wearable device can hear, talk, project data into users’ hands, and presumably onto walls. It can do many of the things our smartphones can, but without the need for a screen.  Retailing for $699 with a $24 per month subscription that can include a phone number and data usage, this is, potentially, huge step in portable capability. But will it catch on? 

🗻 Olympus: Amazon has been working on their own replacement for Alexa to help with shopping, and to add value in the ways other LLMs like GPT do. It will be out for free at the beginning of the New Year.  Notably, Amazon invested $4 billion into Anthropic just two weeks after we announced here that Google has committed $2 billion into the company, which runs Claude/2.  Google already has their own DeepMind, Bert, T5, LaMDA, PaLM, Ernie AI models, so it feels as though Google and Amazon both know something about Anthropic that makes them both very bullish on it. Could a potential power partnership be in the works? Stay tuned!

Written by

Cory Warfield

LinkedIn Top Voice/Influencer in AI

what else?

 🚨 For IE’s daily engineering, science & tech bulletin, subscribe to  The Blueprint  

🧑🏻‍🔧 For expert advice on engineering careers, subscribe to   Engineer Pros   

 🔷 For all the week’s top engineering stories, subscribe to the  Vital Component  

⚙️ To explore the wonders of mechanical engineering, get your  Gears in Motion  

🎬 For a weekly round-up of our best science, tech & engineering videos, subscribe to   IE Originals   

For our weekly premium newsletter and an ad-free experience,   sign up for IE+   

  Give Feedback  

© Copyright 2023 | The Blueprint is by Interesting Engineering,Inc. 530 Fifth Ave, 9th floor New York, NY 10036, USAAll Rights Reserved

You are receiving this email because you have subscribed to our newsletter.Manage your e-mail preferences here.Unsubscribe from our emails here.See our full privacy policy or terms of conditions.