George Davis
@georgebdavis
Followers
258
Following
3K
Statuses
696
CEO @frame_ai & CMU PhD helping machines understand and assist conversations between real people. Let's make systems and businesses as human(e) as possible.
Manhattan, NY
Joined December 2010
Let's spend less time working and talking about AI, and a lot more more on IA and II:
1
0
0
@zacharylipton I'm impressed at how Microsoft has played both sides of this wave -- driving high-parameter obsession by funding OpenAI, while releasing awesome research on distilling, quantization and small models in general.
0
0
0
@largedatabank Love this - any hot takes on what makes this problem different with CockroachDB architecture?
1
0
0
@nntaleb Duty seems to me as hard to purify as altruism. How would you differentiate deontic actions from seeking a pleasure in the fulfillment of duty, or avoidance of the pain felt in dishonor?
0
0
0
@mimurchison Even anthropocentric intelligence should be considered in a task-specific and context specific way. Highly recommend Gould's Mismeasure of Man on that score (and possibly a source of ideas about how we will screw up evaluating AI)
1
0
1
@dauber Reminds me of the y2k masterpiece Romeo Must Die . If you want ownership in an NFL team, you need to make sure Jet Li falls in love with your daughter and protects critical transactions from Triad interference.
0
0
0
@mattturck That's why real VCs go to sleep at midnight in Paris and wake up at 7a Tokyo time for meditation and workout before the opening bell.
0
0
1
@karpathy Love to see this, Andrej - could have an enormous positive impact. Would love to share (with you or anyone else building over there) some of what we learned about both adaptive ed and wrangling the higher ed publishing complex while building Knewton.
0
0
0
@ID_AA_Carmack @fuglypirate Exactly to your point, even within LLMs there is so much valuable research and engineering to be done with open weight models below frontier scale.
0
0
4
AI Management is an under considered issue for enterprise deployment. GAI's failure modes demand monitoring and tuning. For virtual agents, this will look like personnel management. For proactive systems like Frame AI, it means instrumenting efficient feedback loops.
I gave a keynote today about AI Agents. Here's what it boiled down to... 1. Despite university-level intelligence being readily available at our fingertips, production adoption of generative AI in the enterprise remains relatively low. 2. The number one use case for generative AI in the enterprise - customer service - can help us understand why. 3. As the CEO and cofounder of Ada, I've had a front row seat into what's holding back enterprises from deploying generative AI in production: a fear of losing control. 4. AI Management - the practice of measuring, coaching and testing your AI Agents - is the solution to overcoming this corporate fear. 5. Over the last 5 years, the world's top quartile companies in terms of shareholder returns have shared one thing in common: they exhibit the strongest *people * management practices. 6. Over the next 5 years, these companies will also exhibit the strongest management practices with regards to their *non-human* employees too. 7. As AI Agents enter the workforce, it will have profound consequences for the nature of work, one of which being that I think the role of the individual contributor will disappear. Everyone will be expected to be an AI Manager.
0
0
5
For anyone raising a toddler (or a dog!) it is obvious that thought precedes language -- we can study it through the relation between inputs and decisions, even if AGI occultists will call this a black box. But in defense of language, it is an ancient technology for turning thought into a shared experience. It clearly contains many useful biases (alongside chauvinistic ones) regarding a structure for thought -- and specifically a structure of thought designed to interface with and act upon the thinking world.
0
0
2
@mimurchison However, I think Agent-Agent communication may do the opposite -- we will see machines reinvent ambiguity and other forms of deception for strategic comms.
0
0
2
RT @mimurchison: It’s interesting to consider human language as an inefficient technology. What happens when one can bypass linguistic en…
0
2
0
@mimurchison 💯!! With the caveat that "CX-related" is broad - not all these new roles will be in traditional Service orgs. AI makes data and processes more mobile, so we will see more and more functions using customer interaction (and the resulting data) in their core work loop.
0
0
1
@maccaw Hate to hear this. I think in 10y we'll call this era of tech feudal -- by putting cryptic gates on access to their markets, platform holders create a non-fungible form of power to exchange with each other.
0
0
0