The White House lays down the law on AI

The White House lays down the law on AI

Welcome to the third issue of AI Logs, the AI newsletter succinctly highlighting new AI tools, ways to use them, and all things AI. 

My name is Cory Lopes-Warfield, I’ve been a tech founder in the ML and web3 worlds for years, and play with hundreds of tools to bring you the ones I find the most novel, beneficial, disruptive and cool. I appreciate you being a subscriber, and always welcome feedback on how we can make AI Logs the most valuable for you, our esteemed and valued reader.

This week, I have to start by mentioning new AI legislation from the White House in the form of an executive order, that requires:

  • Transparency in reporting data on safety tests

  • Correcting algorithmic biases and prejudices

  • Citizen’s privacy

  • Safeguards against AI being used to create dangerous biological materials

  • Antimonopoly measures and standardization

  • Protection from fraud perpetrated by AI

All were highlighted as priorities. 

There is also a provision for ensuring employers don’t undercompensate workers, but I was struck by the lack of attention paid to both AI taking many (most?) jobs from humans in the not-so-distant future, and the jump to AGI (general artificial intelligence). 

Equally notable, OpenAI just overhauled their core values, mission statement et al to triple down on AGI as their sole focus, and CEO Sam Altman has stated that AGI is not far off. 

For reference AGI can pass the “Turing Test”, meaning a human will not be able to determine if outputs are AI or human derived a “significant fraction of the time”.

And with all that heavy lifting now out of the way, I’m pleased to present Issue 3 of AI Logs

NEWS

⚠️ President Biden to execute the strongest global action on AI safety

The White House recently enacted a sweeping executive order to increase oversight and accountability as AI rapidly progresses. Issued shortly before the UK hosts the first global summit on AI safety, the order establishes new standards for companies to share testing data and mitigate risks.

While representing a bold assertion of US leadership in steering AI responsibly, some question if the unilateral action aligns with the summit’s inclusive tone. 

In contrast, the UK is seeking greater global cooperation and aims to engage nations at all levels of AI development. But the US order is framed more narrowly through an “American leadership abroad” lens.

The urgency is understandable as AI influence surges exponentially. But truly global governance requires looking beyond major players like the US and China. So while the order has landmark significance, it could undermine the collaborative spirit of the summit. The ideal path forward likely rests between wide international coordination and decisive domestic action by major AI powers.

Click below to explore the implications of this US order on AI safety and the complexities of balancing competition and cooperation globally.

Other Important Updates

Prompt of The Week

This week, I’ll showcase two short prompts I love to use that have been game changers for me. 

✍️ The first is asking your preferred AI model to prompt itself by first providing context and prompts of your own, and then asking it to act as an AI expert prompt engineer and rewrite the prompts you already wrote in a manner sufficient to get better results/output. 

You can ask it to explain why its prompts are superior to understand the logic behind them, and then reuse the generated prompts to take your initial task(s) to the next level! 

✋ The other is simply telling the model what not to do, and my favorite use-case is to tell it it doesn’t need to ask permission to ‘proceed’ ‘get started’ et al in between prompts, but rather can act autonomously, can make assumptions (so it doesn’t keep asking you a million questions; but always “check its math”), and effectively “prompt itself” (autonomous agent). 

This is most powerful when coupled with the ‘Professor Synapse’ super-prompt I provided in Issue 1.

👍 Bonus quick prompt hack: if you’re ever stuck in GPT where it says it’s doing what you asked but is really doing nothing, just type the letter Y and send it as a prompt. It will then actually do what you asked it to!

Create your own visual universe with EverArt

Did you know there’s now a tool that enables you to generate art in your own distinctive style, or to take an existing style and create endless re-interpretations of scenes and characters? 

That's what Nick St. Pierre trained d to produce Taylor Swift and Larry David in the style of Rick Morty , as shown here. It's called EverArt , and it could soon enable you to “turn your unique style into an infinite canvas for creation” too.

TUTORIAL

ChatGPT Pro Vision

Last week I went over how to use DALL•E 3 in ChatGPT Pro. To recap:

  1. Toggle “Beta features” on 

  2. Select DALL•E 3 modality (although I can’t wait until Pro users can use Browse with Bing, Advanced data analytics,DALL•E 3, and Plugins simultaneously)

👀 Today, I’ll highlight “the other” big upgrade to ChatGPT Pro: Vision. You can now take or upload a picture and chat with ChatGPT about it. 

A few examples to get the creative juices flowing include taking a picture of your houseplant and asking “What’s that yellow spot?”, “OK, how do I provide the right nutrients to make that go away?”, or “what else can you tell me about the plant and it’s over-all health?”

Another example would be to take a picture of your car’s engine and ask questions about its performance, what the rattling sound could be, how to change the oil, etc. 

This Vision tool is powerful, and it can be further enhanced within ChatGPT’s official mobile app with speech and listening capabilities (like Siri, Alexa, etc but with better voices and obviously way better output).

TOOLS OF THE WEEK

🗣 Coachvox is an incredible AI that you can use to “clone yourself” and train on your own data - in other words, upload files and videos of yourself, set up how it should interact with others, and then very easily deploy “your own AI” that can coach people, to engage your audience, and much more. It’s definitely skating to where the puck is going, an a giant step towards personal digital twins. 

👩‍💻 Anthropic/Claude 2 (who just received $2 billion committed investment from Google!) now has VSCode - CodeGPT Copilot available in V 2.2.1 that is powerful at code auto completion for those using AI to write/generate code or expedite programming projects.  CodeGPT also works with ChatGPT — who’s Advanced Data Analytics is pretty good at writing code from scratch (particularly in Python) — and Google PaLM2. It will be interesting to see what happens with Bard and PaLM2 now that Google has invested so heavily in Anthropic.

💬 Meta AI’s AI chatbots have been available for a few weeks as limited celebrities (like Snoop Dogg as a Dungeon Master to help gamers), the implication is massive as they’re suggesting that soon average users may be able to create avatars of themselves trained on all the data we’ve provided them with for years now.  This could be a game changer, and I definitely wouldn’t sleep on Meta, Apple, and Amazon’s AI programs and products even if the current leaders in the space appear to be Google and Microsoft (via OpenAI) at this stage.

🎨 Canva is rolling out more AI capabilities such as Magic Write and Text-to-Image, and also has a great plugin for ChatGPT that can find the right templates and schemas and suggest all the plug-and-play content to go into the presentation, although the next logical iteration will be text-to-presentation directly inside of both ChatGPT and Canva.  Canva has been an early adopter of incorporating AI into their amazing platform. Most entrepreneurs, solopreneurs, creatives, executives and professionals I know — myself included — always list Canva as one of, if not their very top tool to use. I use it daily and can’t wait to see more AI integration and capability. 

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+   

© 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.