- AI For All
- Posts
- OpenAI Prepares for AGI
OpenAI Prepares for AGI
PLUS: Navigating Horizontal and Vertical AI Solutions
AI For All
December 22, 2023
Hello readers,
Welcome to the last edition of 2023! OpenAI has released a Preparedness Framework for ensuring that next-gen models are safe to deploy. Plus, learn about navigating horizontal and vertical AI solutions from a global supply chain company.
As always, thanks for being a subscriber! We hope you enjoy this week’s content — for a video breakdown, check out the episode on YouTube.
Let’s get into it!
Do You Feel the AGI?
Image generated by DALL·E 3
OpenAI has established a Preparedness Framework for frontier models, one of which may eventually be artificial general intelligence (AGI). The Preparedness team, led by Aleksander Madry, focuses on the spectrum of risks associated with AI advancements. These risks span from the immediate, like the potential for AI models to be used in phishing attacks, to more far-reaching and speculative threats, including chemical, biological, radiological, and nuclear dangers.
The Preparedness team is part of a broader initiative within OpenAI that encompasses various safety teams. For instance, the Safety Systems team handles current model misuse, while the Superalignment Team focuses on future superintelligent models.
A key component of the Preparedness team's strategy is the development of a risk-informed development policy. This policy is designed to guide the evaluation and monitoring of AI models, informing risk mitigation actions and governance structures throughout the model development process.
Moreover, OpenAI has launched the AI Preparedness Challenge to engage the broader community in this effort. The challenge invites individuals to propose ideas for risk studies, with the top submissions being rewarded $25,000 each in API credits and potentially offered a role on the Preparedness team.
Our Take
Do you feel the AGI? A question OpenAI Chief Scientist Ilya Sutskever reportedly asked his coworkers constantly, even leading them in a ritualistic chant: “Feel the AGI!” With all this preparation, one might suspect something is brewing at OpenAI, or it’s just par for the course safety research, perhaps to restore faith in the company after the CEO drama. Notably, the OpenAI board now has the power to veto the release of new models. We’ve always welcomed safety research to ensure that AI is beneficial for individuals and businesses. As 2023 was unequivocally the craziest year for AI, we expect things to stabilize in 2024 (or not). Whatever happens, we’ll be sure to tell you about it.
🔥 Rapid Fire
Google unveils VideoPoet for video generation
Apple releases two new AI research papers
Google shares plans to mitigate US election misinformation
Microsoft Copilot adds song generation with Suno
Stability AI announces paid membership
MIT researchers discover new antibiotic candidates with AI
Anthropic will defend customers from copyright claims
EU expands support for AI startups
UK Supreme Court rules AI is not an inventor
IBM acquires tech from Software AG to bolster AI offerings
BrainChip previews new edge AI device
Expedia plans to add AI to cut out Google
ByteDance allegedly uses OpenAI API to build competitor
AI companies collaborate to end hallucinations
Pentagon is being urged to better define its AI workforce
🎙️ The AI For All Podcast
This week’s episode featured Andrew Reiner, the co-founder and CEO of Riley, who discussed the landscape, challenges, and strategies for AI startups and businesses. We dived into the transition from predictive modeling to impactful AI solutions in startups, the need to focus on major operational problems to build successful AI products, and advice for building AI startups and businesses.
📖 What We’re Reading
“In the age of artificial intelligence (AI), the choice between horizontal and vertical development models has become a crucial factor in determining how effective AI solutions are across different industries. This strategic decision is pivotal as horizontal AI aims to address a wide range of challenges across sectors, while vertical AI takes a more focused approach, tailoring solutions to meet the specific needs of particular industries.”
8 Myths Keeping Manufacturers From Innovating At Scale (link)
“AI innovation in industrial manufacturing is needed to solve complex business problems. Yet, managing the change that accompanies the innovation is complex. But once you’ve busted through some common myths and found the right partners, the process becomes much easier.”