The 4 Most Important ChatGPT News Items Last Week (Mar 25) in 90 Sec

Tons of product announcements. ChatGPT wins security contest. AGI, you're kidding, right?

1: Google launched Bard (ChatGPT competitor) and ChatGPT releases Plugins

Why it matters: Basic text-to-text Generative AI has, stunningly, become a commodity with offerings now from Google, Microsoft, OpenAI/ChatGPT, Claude, Poe and others. Roughly 4.9 billion people use Google search whereas Microsoft may have only 300M Office users and (10M? Bing users). Google will integrate Bard deeply with its search and be the market leader in a year for basic text-to-text Generative AI, and ChatGPT will become more specialized (APIs for developers and an incredible tool for creative/prose tasks). Here is our initial comparison the ChatGPT vs Google’s Bard and Microsoft Bing AI.

2: New artificial Generative AI capabilities were announced by dozens of firms including Microsoft (text to image), Adobe (Firefly for images), Runway (text to VIDEO), Canva, etc.

Why it matters: Yes, vendor claims are flooding the airways. Just because a product can integrate with ChatGPT, that doesn’t mean that the integration is a great use. Expect the ChatGPT or Bard “prompt’ or search bar to be accessed within the UX from any personal or business apps. Generative AI at our fingertips within our favorite app for creating text and images.

3: ChatGPT entrant wins security hackathon. A team from cybersecurity firm Claroty used ChatGPT to write code to exploit vulnerabilities in IT systems.

Why it matters: Using ChatGPT to write kids stories at night is cute. Using ChatGPT to hack into corporate and industrial companies is not. The ability to learn and adjust at scale (“create 10M different security (digital and phishing) attacks for Department of Defense. For each failed attempt, create a new attack based on what failed. Keep repeating this until file is uploaded to the internal network”.

4: Microsoft Research’s 154p report concludes that GPT-4 is an every version Artificial General Intelligence (AGI). What?? You are kidding, right?

Why it matters: Much of the AI breakthroughs have been on complex but narrow use cases (win at Chess or Go, complete protein folding (yes amazing), optimize search ads for clicks, speech recognition, medical X-ray analysis, etc.), but the dream by some was artificial general intelligence, a broad and horizontal capability, that displays an expansive mental capability to reason, plan, solve problems, think abstractly, comprehend complex ideas, learn quickly, learn from experience, respond based on a Theory of Mind, etc. The research paper is somber and full of mathematical tests. Read the first 5 pages and determine your own view on this topic.

Friends of ChatGPTNuggets doing cool things:

• Adam Rappaport in Australia has “100 days of ChatGPT” on Twitter, (he recently finished 50)

• Simon Greenmail in the UK wrote a thoughtful piece about “Who will make money the Generative AI gold rush?

• Charles Blanchet in France is focused on helping French legal firms gain value from ChatGPT

• Dave Edwards at Sonder Studio is doing some great ChatGPT research

Join us at our weekly ChatGPT learning/Jam session (signup on Meetup), it’s the largest Meetup group in the U.S. focused on ChatGPT. Tonight will include demos of GPT-4, Google’s Bard, Microsoft’s BingAI, and ChatGPT Plugins.

"Time and space are not conditions in which we live, but modes by which we think.” - Albert Einstein

Onward,
Paul

P.S. If you enjoy this newsletter, consider suggesting to others that they subscribe here.