StarZero All shows Try the podcast clipper →

Jul 28, 2017 · 23m · a16z

Mark Ring

Mark Ring · 19m spoken
0:00 / 0:00
▶ Watch on YouTube →

gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions

At the Andreessen Horowitz Academic Roundtable, Mark Ring introduces continual learning and isolaminar forecast mechanisms as the fundamental pathway to true artificial intelligence. Through a simulated robot thought experiment, he demonstrates how high-level spatial abstractions can be constructed incrementally from raw sensorimotor data.

How this conversation actually went

Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. How this is scored →

The host as informed peer 0.0 Guest teaching 4.4 Guest disagreement 0.4 The host pushing back 0.0
05100:0010:0020:000:21–2:58 · The host as informed peer 0/10 The Path to AI and Current Limitations In this solo presentation segment, Mark Ring critiques modern AI as a computationally sophisticated party trick while explaining the progression toward continual learning. The host is entirely absent, making host-side scores zero.2:58–7:55 · The host as informed peer 0/10 The Concept of Continual Learning and Isolaminar Mechanisms Ring details the concept of isolaminar learning mechanisms using human developmental analogies such as learning to sit before walking. Host metrics remain at zero due to host silence.7:55–10:33 · The host as informed peer 0/10 Microworld Thought Experiment Setup Ring introduces a thought experiment involving a simplified robot in a microworld with raw sensory inputs. The guest operates in solo monologue mode without host interaction.10:33–18:58 · The host as informed peer 0/10 Building Abstractions Layer by Layer with Forecasts Ring explains how low-level sensor feedback can be layered with forecasts to construct abstract concepts like distance and walls. The monologue dynamic continues with zero host involvement.18:58–21:49 · The host as informed peer 0/10 Constructing High-Level Spatial Abstractions Ring expands the thought experiment to show how higher-order spatial structures like rooms and houses are represented through forecast networks. Host scores remain zero.0:21–2:58 · Guest teaching 4/10 The Path to AI and Current Limitations In this solo presentation segment, Mark Ring critiques modern AI as a computationally sophisticated party trick while explaining the progression toward continual learning. The host is entirely absent, making host-side scores zero.2:58–7:55 · Guest teaching 5/10 The Concept of Continual Learning and Isolaminar Mechanisms Ring details the concept of isolaminar learning mechanisms using human developmental analogies such as learning to sit before walking. Host metrics remain at zero due to host silence.7:55–10:33 · Guest teaching 4/10 Microworld Thought Experiment Setup Ring introduces a thought experiment involving a simplified robot in a microworld with raw sensory inputs. The guest operates in solo monologue mode without host interaction.10:33–18:58 · Guest teaching 5/10 Building Abstractions Layer by Layer with Forecasts Ring explains how low-level sensor feedback can be layered with forecasts to construct abstract concepts like distance and walls. The monologue dynamic continues with zero host involvement.18:58–21:49 · Guest teaching 4/10 Constructing High-Level Spatial Abstractions Ring expands the thought experiment to show how higher-order spatial structures like rooms and houses are represented through forecast networks. Host scores remain zero.0:21–2:58 · Guest disagreement 2/10 The Path to AI and Current Limitations In this solo presentation segment, Mark Ring critiques modern AI as a computationally sophisticated party trick while explaining the progression toward continual learning. The host is entirely absent, making host-side scores zero.2:58–7:55 · Guest disagreement 0/10 The Concept of Continual Learning and Isolaminar Mechanisms Ring details the concept of isolaminar learning mechanisms using human developmental analogies such as learning to sit before walking. Host metrics remain at zero due to host silence.7:55–10:33 · Guest disagreement 0/10 Microworld Thought Experiment Setup Ring introduces a thought experiment involving a simplified robot in a microworld with raw sensory inputs. The guest operates in solo monologue mode without host interaction.10:33–18:58 · Guest disagreement 0/10 Building Abstractions Layer by Layer with Forecasts Ring explains how low-level sensor feedback can be layered with forecasts to construct abstract concepts like distance and walls. The monologue dynamic continues with zero host involvement.18:58–21:49 · Guest disagreement 0/10 Constructing High-Level Spatial Abstractions Ring expands the thought experiment to show how higher-order spatial structures like rooms and houses are represented through forecast networks. Host scores remain zero.0:21–2:58 · The host pushing back 0/10 The Path to AI and Current Limitations In this solo presentation segment, Mark Ring critiques modern AI as a computationally sophisticated party trick while explaining the progression toward continual learning. The host is entirely absent, making host-side scores zero.2:58–7:55 · The host pushing back 0/10 The Concept of Continual Learning and Isolaminar Mechanisms Ring details the concept of isolaminar learning mechanisms using human developmental analogies such as learning to sit before walking. Host metrics remain at zero due to host silence.7:55–10:33 · The host pushing back 0/10 Microworld Thought Experiment Setup Ring introduces a thought experiment involving a simplified robot in a microworld with raw sensory inputs. The guest operates in solo monologue mode without host interaction.10:33–18:58 · The host pushing back 0/10 Building Abstractions Layer by Layer with Forecasts Ring explains how low-level sensor feedback can be layered with forecasts to construct abstract concepts like distance and walls. The monologue dynamic continues with zero host involvement.18:58–21:49 · The host pushing back 0/10 Constructing High-Level Spatial Abstractions Ring expands the thought experiment to show how higher-order spatial structures like rooms and houses are represented through forecast networks. Host scores remain zero.

speaking balance: gold is the host, purple is the guest (3 minute bins)

0:00 · the host 0% · guest 100%0:00 · the host 0% · guest 100%3:00 · the host 0% · guest 100%3:00 · the host 0% · guest 100%6:00 · the host 0% · guest 100%6:00 · the host 0% · guest 100%9:00 · the host 0% · guest 100%9:00 · the host 0% · guest 100%12:00 · the host 0% · guest 100%12:00 · the host 0% · guest 100%15:00 · the host 0% · guest 100%15:00 · the host 0% · guest 100%18:00 · the host 0% · guest 100%18:00 · the host 0% · guest 100%21:00 · the host 0% · guest 100%21:00 · the host 0% · guest 100%
Sharpest disagreement ▶ 0:21 Dismissing modern AI as a party trick

Mark Ring mildly challenges conventional industry optimism by dismissing current AI tools like Siri as computationally sophisticated party tricks.

 Hardest push from the host ▶ 0:21 Absence of host pushback

The segment is a continuous presentation where the host offers no pushback or framing challenges.

 Biggest teaching moment ▶ 14:50 Grounding abstract concepts in sensory forecasts

Ring clearly demonstrates how low-level sensory motor streams can be mathematically built up into abstract concepts like walls without manual human labeling.

 The host holds their own ▶ 0:21 Absence of host counter-expertise

Because the transcript is a presentation monologue, the host does not intervene to assert expertise or challenge the speaker.
the scores for every segment, with the reasoning behind each
ChapterTopicThe host as informed peerGuest teachingGuest disagreementThe host pushing backWhy
The Path to AI and Current Limitations 0420 In this solo presentation segment, Mark Ring critiques modern AI as a computationally sophisticated party trick while explaining the progression toward continual learning. The host is entirely absent, making host-side scores zero.
The Concept of Continual Learning and Isolaminar Mechanisms 0500 Ring details the concept of isolaminar learning mechanisms using human developmental analogies such as learning to sit before walking. Host metrics remain at zero due to host silence.
Microworld Thought Experiment Setup 0400 Ring introduces a thought experiment involving a simplified robot in a microworld with raw sensory inputs. The guest operates in solo monologue mode without host interaction.
Building Abstractions Layer by Layer with Forecasts 0500 Ring explains how low-level sensor feedback can be layered with forecasts to construct abstract concepts like distance and walls. The monologue dynamic continues with zero host involvement.
Constructing High-Level Spatial Abstractions 0400 Ring expands the thought experiment to show how higher-order spatial structures like rooms and houses are represented through forecast networks. Host scores remain zero.

Statements from this episode (10)

Opinion
Mark Ring: A lot of modern AI is just a sophisticated party trick
“To a large extent, I think, ah, a lot of AI today is really a computationally sophisticated party trick.”
Mark Ring Jul 28, 2017 ▶ 0:25
Assertion Not checkable as stated
Mark Ring: AI cannot answer questions requiring world visualization
“Nearly any question that requires visualizing how the world works will go well beyond the ability of any AI technology in existence today.”
Mark Ring Jul 28, 2017 ▶ 1:00
Opinion
Mark Ring: Continual learning is the only path to real AI
“Continual learning is about building machines that learn on their own, which I think is the best, maybe the only path to real AI.”
Mark Ring Jul 28, 2017 ▶ 1:53
Assertion Not checkable as stated
Mark Ring asserts the human brain is an isolaminar learning mechanism
“The brain Is an isolaminar mechanism. You get one brain. No one gets to tinker with it and change it while you're growing up.”
Mark Ring Jul 28, 2017 ▶ 3:59
Insight
Ring: AI forecasts encode knowledge as subjective, behavior-dependent predictions
“Their secret is that they encode knowledge as behavior-dependent predictions, meaning that each forecast predicts something about what an agent will perceive if it acts in a certain way. So they're subjective predictions. Each one says basically, I will percei…”
Mark Ring Jul 28, 2017 ▶ 5:23
Insight
Ring: AI forecasts enable layered knowledge by predicting other forecasts
“But critically, forecasts are an isolaminar mechanism, so you can use them layer after layer, from simple to complex. And this is possible because a forecast can make a prediction about another forecast. You can build forecast A to make one prediction, and the…”
Mark Ring Jul 28, 2017 ▶ 5:59
Opinion
Ring: Continual forecast chains show first glimpse of true AI understanding
“So I think that what we're seeing here is a first glimpse of true computer understanding of the world around us in the way that, that humans have it.”
Mark Ring Jul 28, 2017 ▶ 17:41
Insight
Ring: True robot knowledge stems from experience, not pre-programmed labels
“There, these aren't just arbitrary labels or categories that are programmers assigned to a node in a graph. This is verifiable knowledge based on experience in the real world.”
Mark Ring Jul 28, 2017 ▶ 17:54
Insight
Mark Ring: Predictive forecasts can build endlessly abstract AI representations
“Distance maps to doorways, hallways, rooms of different shapes and sizes, and in principle, any distinctive set of forecasts can be predicted, and those predictions can be used to build ever more abstract forecasts continuing indefinitely.”
Mark Ring Jul 28, 2017 ▶ 19:28
Prediction Not checkable as stated
Ring: Technical hurdles in AI forecast mechanisms will be solved within years
“These are all relatively soluble problems, and nothing would lead me to believe that they can't be solved within the next few years given sufficient focus and effort.”
Mark Ring Jul 28, 2017 ▶ 21:57
Made with StarZero

Turn any episode into a week of clips.

This entire site, over 1,000 episodes transcribed, diarized, checked and made playable, runs on the StarZero media pipeline. Drop in your own episode and the podcast clipper finds the moments worth sharing, cuts them, captions them, and reframes them for every feed.

Try the podcast clipper Learn more