šĀ On time for your weekend: a round-up ofĀ this weekās remarkable storiesĀ at the intersection of technology, business, design, and culture. Three reads and three listens; no fluff, just stuff ā”
The past is written, but the future is left for us to write. And we have powerful tools, openness, optimism, and the spirit of curiosity āĀ Capt. Jean-Luc Picard
š Reading
AI and Trust:
The natural language interface is critical here. We are primed to think of others who speak our language as people. And we sometimes have trouble thinking of others who speak a different language that way. We make that category error with obvious non-people, like cartoon characters. We will naturally have a ātheory of mindā about any AI we talk with.
Bruce SchneierāSchneier on Security | 17 minutes
What LLMs cannotĀ do:
LLMs in no sense āthinkā like people do [ā¦] I donāt think this point is controversial, but it seems to often be ignored in practice, not least when people start talking about LLMs being āhuman-likeā or indeed āsuperhumanā. We need to stop comparing LLMs to humans, and instead understand and evaluate what they can and cannot do, without assuming that their abilities in any way correlate with human abilities.
Ehud Reiter | 5 minutes
A Camera, Not an Engine:
[M]odern AI doesnāt seem like an invention at all, let alone one embodied by the behavior of a āmachine.ā The word weāre looking for isĀ discovery.Ā To the extent things like racks full of GPUs and PyTorch shuffling data around constitute a āmachineā in the picture, they are like the Webb telescope; instruments of discovery rather than engines of production
Venkatesh RaoāRibbonfarm Studio | 22 minutes
š§ Listening
AI and Everything Else:
Potentially with an LLM, with generative AI, you could just go and tell the computer what you want. So you don't need someone to have created a separate tool for every one of the 500, or 1,000, or 5,000 different tasks. So you can have far less software doing far more. And so you can automate massively more tasks. You can have massively more problems being solved, but actually with fewer tools. Maybe.
Benedict EvansāSlush 2023 | 27 minutes
Intelligence requires a biological brain:
Intelligent beings appear to have a brilliant ability to [ā¦] generate knowledge and becoming Turing complete on our own. Although AIs are built in Turing complete things, their thinking is not Turing complete in that they are not able to build universal explanations and that lack of universal explanation means that they're just inductivists. Inductivism doesn't get you anywhere and it's just basically a party trick.
Lee CronināLex Fridman | 13 minutes
Google Gemini, Meet Crypto-Cyber-Crystals:
The reality is everyone's going to be putting out demo videos and look cooler and cooler. We're just going to kind of roll through a bunch of marketing. And it's going to look super shiny. And it's going to look cooler and cooler. Everyone's like, "Oh, this one's winning. That one's winning." Look, we'll see in five years [ā¦] I'm really curious how the enterprise revenue stacks up, what developers choose.
Sam LessināMore or Less | 53 minutes
š Timeless
1ļøā£ year ago: AI and I: The Age of Artificial Creativity
2ļøā£ years ago: The Toyota Production System: A Love Letter
3ļøā£ years ago: What comes after smartphones?