Me: What is Knowledge-First AI?
GPT: Knowledge-first AI is a type of AI that focuses on using knowledge to solve problems, as opposed to other approaches that focus on learning from data.
Set a reminder to join and listen to industry leaders and investors talk about #IndustrialAI for the $25T Industrial economy.
Friday 11/11 at 7pm PST!
x.com/i/spaces/1BdGY…
GDP explains almost 70% of pessimism toward AI. But perhaps not in the direction you think.
@StanfordHAI has come out with the excellent annual AI Index aiindex.stanford.edu/report/. It referenced an IPSOS Global Survey from 2022 on positive/negative expectations from AI.
The whole clip shows @karaswisher being intellectually honest about Elon, seeing all the good, bad, and ugly. And that Bill Maher agrees with her assessments.
Is that not better than to be a simpleton fanboy/hater?
“This week on TheDataExchangePod: @pentagoniac, CEO @aitomatic. We discuss the unique challenges and opportunities in combining human & machine intelligence.” HT @bigdata.
They can absolutely both be true, or rather, best estimates. This is because the numbers come from entirely different sets (see Venn diagram below): (A) company payrolls (“Establishment”, YELLOW), which does not include a whole lot of jobs that *are* included in (B) individual
By this measure, search was not “free” either, Naveen. Google did benefit quite a bit from the ’00 crash with capex, and “cheated” for a while with power until one of our racks burned down a colo 😎. And you’re right that a profitable business model took time to find—long-tail
I love your objective insights @balajis, but this perspective is too us-centered. The better way to understand the Chinese perspective is that we obsess about them A LOT more than they think about us.
Chinese companies and technologists have always been all-in on open source
12 years ago, a colleague from SRI (“Siri”) & I struggled for months—to build a pipeline of audio codecs, speech & language models out of bi-grams—to build a marginally acceptable ASR for a mobile-butler app.
Today something 100x better can be done in dozens of lines of code.
New tutorial on keras.io: build an automatic speech recognition system using a CTC loss and a model inspired by DeepSpeech2.
End-to-end (from raw wave files to inference with the trained model) in... *200 lines of code** 🤯keras.io/examples/audio…