AI Top-of-Mind for 5.7.24 — What is MAI-1?
Top-of-mind is the rumored new Microsoft model, MAI-1. Its training data is vastly larger than the company’s open source models and is designed to go up against the likes of GPT-4. It is not based on anything from OpenAI and takes minimal technology from ‘Inflection.’ More from ‘The Information’ and from ‘Ars Technica.’
Also on models, and the use of ‘agents.’ Han Heloir writing in ‘Towards Data Science’ offers a great tutorial into how they operate across various platforms like AWS and Gemini. She provides a good framework:
But with AI models come bias, and some good news from Morgan State University’s Center for Equitable Artificial Intelligence and Machine Learning Systems (CEAMLS). As the ‘EE Times’ article relates:
· CEAMLS devised BiasWatch, an online platform to collect anecdotal evidence of bias from the public that’s intended to help inform decisions when choosing a large-language model (LLM). Plans are to automate the process at some future point.
· …and is testing the LLM Latimer, commonly referred to as the “Black ChatGPT.”
· Latimer is built to more accurately reflect the experience, culture and history of Black and brown people. Its foundation is ChatGPT, one of the cloud-based tools that many LLMs use as their base, and are of concern due to inherent bias
And deepfakes are still a concern. ‘SmartBrief’ looks at how AI is aiding phishing, including voice cloning. From the article:
· The classic phishing email has gotten an AI boost as well, as generative AI can produce text without the spelling and grammatical errors that frequently give away human-authored attempts at gaining access.
· The FTC has started trying to tackle AI in its anti-fraud efforts, with a new contest aimed at helping people recognize voice clones, the ability to sue entities that impersonate government agencies and a proposed rule that would make companies liable if their AI tools are used in deepfake scams. AI-based fraud probably isn’t going anywhere, though, even if that proposal becomes effective: businesses and consumers alike must stay alert and double check anything that seems suspicious, even if it ostensibly comes from their grandchildren or Taylor Swift.
On to marketing, and the promise of AI for digital agencies. The ‘Campaign US’ article looks at Gen AI as transformative and will counter budget cutbacks that agencies are experiencing. From the article:
Digital agencies, designed for innovation, are the natural candidates to guide marketers through to this shift. They can add value here by helping brands prepare for a world where customer interactions with their AI experience is more front and center than how they show up in their advertising.
Closely related is retail, where Zack Kass, interviewed in ‘Forbes,’ states that AI will help retailers survive. From the article:
· “I believe the single most important factor for companies successfully working with AI is an executive team committed to its potential. Some boardrooms and executives in Fortune 100 companies are not yet convinced that AI will revolutionize their businesses or industries. It doesn’t matter what else happens within these companies; they will likely struggle.”
· “So, connecting dots — humans are still required for this. But the major piece that we observe in AI is that it actually is not capable of doing the things that require wisdom, courage, vision, empathy, and curiosity; these immutable human qualities.”
And speaking of human skillsets going forward, Axios published a report on AI hiring across the US. No surprises for the known ‘AI hubs,’ but some of the secondary locations are surprising.
Lastly, the future of war… If you remember the movie ‘Maverick,’ a key theme was the replacement of manned fighters by unmanned drones. From the ‘Mercury News’ article:
· With the midday sun blazing, an experimental orange and white F-16 fighter jet launched with a familiar roar that is a hallmark of U.S. airpower. But the aerial combat that followed was unlike any other: This F-16 was controlled by artificial intelligence, not a human pilot. And riding in the front seat was Air Force Secretary Frank Kendall.
· AI marks one of the biggest advances in military aviation since the introduction of stealth in the early 1990s, and the Air Force has aggressively leaned in. Even though the technology is not fully developed, the service is planning for an AI-enabled fleet of more than 1,000 unmanned warplanes, the first of them operating by 2028.