-
Why You Need a Thinking Partner (like this Genius)
You may not have heard of him, but James Clerk Maxwell wrote down what we now think of as four equations that define ALL of electromagnetics. You can get a T-Shirt with them emblazoned on it. If you want to know how radio waves propagate, or how a radio functions, or how anything running on electricity…
-
A change of perspective is worth 80 IQ points
We’ve all done it–struggled over some problem until a frustrating end of the day. But after “sleeping on it” and stepping into a hot shower, a breakthrough idea pops up, as if out of nowhere. This often comes from getting too close to the problem. A little distance can make a decisive difference. The quotation…
-
Teaching an AI to use a knowledge graph for semantic compression
The following conversation has implications for MemGPT and similar use cases, as well as LLM/Knowledge Graph integration. Instead of requiring URIs for everything (which would require constant lookups against a lexicon of all-possible-things) this uses locally non-ambiguous (aka conversational) identifiers. Something like this would be great for tracking continuity in fiction, for instance. Maybe a…
-
New LinkedIn Learning course on building AI applications in development
I’ve signed a contract with LinkedIn Learning to record a course on building AI apps. Exact title TBD. OpenAI’s announcements this week certainly make things interesting. This whole space is changing so quickly that figuring out exactly what to cover is its own serious challenge. What are your most burning questions about building AI apps…
-
The AI of the Gaps Argument: What Most People Miss
Modern AI technologies have from their origin in the 1950s have been characterized by what that can’t do. This doesn’t look like a coincidence, and it has long-term implications on important work toward regulating, legislating, and leaning to live with AI technologies. It wasn’t so long ago that serious-minded folks could be found claiming that…
-
Wolfram Alpha and ChatGPT make a power couple
Originally posted on LinkedIn in March, 2023 It’s happening. Earlier this year I opined about how AI systems would interface with authoritative data sources, based on a Stephen Wolfram blog. Insert your own comments here about the pace of change…but, yeah, it happened. And SW is on it. https://writings.stephenwolfram.com/2023/03/chatgpt-gets-its-wolfram-superpowers/ This is good news for systems…
-
Why talking it out helps solve problems: Leveraging latent spaces
I’m someone who benefits a lot from talking out a problem–often in front of a whiteboard. This is pretty common. But why does it work? We had been previously discussing latent spaces — a kind of internal map connecting different pieces of knowledge in different ways — as part of “understanding” and this is another…
-
Does Generative AI Understand? Surprisingly: It kind of does
Dmitri Mendeleev was a smart guy. In 1869 he wrote out a card for each element and started sorting them by increasing atomic weight. He noticed that there were repeating patterns, for instance, the noble gases. Thus was forged the periodic table of the elements. Question: Did Mendeleev “understand” the underlying structure in a way…
-
LinkedIn Layoffs: free mentoring for affected engineers and managers
I’m sorry to hear about all the layoffs today. I’ve been through that more than once in my career, and it is crushing. Once the shock wears off, it just feels…depressing. Please take some time and digest the situation, and reflect on your many accomplishments and the many lives who have been brightened by your…
-
Will LLMs always take millions of dollars to train? Three factors to consider
Estimated costs to train GPT-4 range from $60 to $100 million USD. But will it always take millions of dollars to train a new foundation language model? There are (at least) three different forces in play, and they are pulling in different directions. (1) The predominant factor is “bigger is better”. The more parameters in…
-
Will AI Oversaturate? (2000s Calling… They Want Their Hype Cycle Back)
If the “XML hype cycle” doesn’t spark vivid memories, go ask someone who’s been in IT for 25 years. 🙂 XML was to be the solution to all of the world’s data problems. C-Suite execs who wouldn’t know a parser from a parsnip we getting routinely quoted in press releases, echoing the praises of their…
-
Hollywood Writers Strike Ends with Surprisingly Reasonable AI Agreement
It was the longest strike to date, but now it’s over. Here’s a pretty good summary of the agreement. Of course, I like to think that the withering pressure of the threat to replace studio execs with AI had something to do with it. 🙂 As far as AI goes, the basic agreement seems reasonable:…