Positive transfer learning will continue to emerge, meaning more diverse data makes Figure robots broadly better at many tasks.
Predictor: Brett Adcock · ep#229 "The Humanoid Takeover: $50T Market, Figure's Full Body Autonomy, and Robots in Dorms #229" · source
Prediction text
Positive transfer learning will continue to emerge, meaning more diverse data makes Figure robots broadly better at many tasks. | we've already shown positive transfer now with all this data. Like coming in the robot can journalize better with more information even like more knowledge is better... I hope this will lead to really um uh I think you'll see a lot of positive transfer emerge from a robot that's able to like generalized to a lot of things.
Watch events: Figure 03 shipment volume; BMW, 2nd customer deployment count
Verbatim quote
we've already shown positive transfer now with all this data. Like coming in the robot can journalize better with more information even like more knowledge is better... I hope this will lead to really um uh I think you'll see a lot of positive transfer emerge from a robot that's able to like generalized to a lot of things.
Predictor: Brett Adcock
Calibration plot (stated vs observed)
Evidence about this node from Brett Adcock is multiplied by κ in /api/intake. Lower κ = less weight; floors at 0.10 (effectively silenced) and caps at 1.00 (full weight).
Reference class: humanoid_commercial_volume
>10,000 unit cumulative deployment of humanoid robot SKU within 3 years of debut
Tetlock-style outside view: at TRF=1 (just predicted), outside view dominates (w_in=0.3). At TRF=0 (deadline), inside view dominates (w_in=1.0). The blend regularizes overconfident inside views toward the historical base rate.
Probability over time
Milestone chain
- 2026-09-18pendingQ1 window check-in (25%)
- 2026-04-30partialFigure publishes positive-transfer benchmark showing diverse-data → broader task generalizationHow: Figure / Helix research note quantifies that adding non-target-task data improves performance on held-out tasks (positive transfer coefficient > 1.0 across domains)Source: Figure already demonstrated zero-shot human-to-robot transfer with Project Go-Big using egocentric human videoconf 85%Notes: PARTIAL — Project Go-Big and Helix-02 already demonstrate cross-embodiment positive transfer (human video → robot navigation). Quantified benchmark across many tasks pending.
- 2027-02-06pendingQ2 window check-in (50%)
- 2026-06-01 → 2027-12-31pendingHelix 02 (or successor) demonstrates room-scale autonomy on novel tasks not in training distributionHow: Figure releases technical report or video of Helix model executing 30+ unseen home/warehouse tasks zero-shot from natural languageSource: Helix 02 Full-Body Autonomy announcement; Adcock's 'end of C++' framingconf 65%
- 2026-06-01 → 2028-06-30pendingIndustry-wide acceptance of VLA (Vision-Language-Action) models as standard humanoid stackHow: Tesla, 1X, Unitree, Agility all ship VLA-based stacks; transformer-based end-to-end policies become default robotics curriculumSource: State of Robotics 2026 already calls VLA adoption the dominant 2026 trendconf 80%
- 2027-06-27pendingQ3 window check-in (75%)
- 2026-09-01 → 2028-06-30pendingCross-domain transfer: warehouse Helix model generalizes to home tasks (or vice versa)How: Same Helix checkpoint demonstrably executes tasks across logistics + household + manufacturing without per-domain finetuningSource: Helix Logistics deployment + home Brookfield pilots running parallel — natural test of cross-domain transferconf 50%
- 2026-12-01 → 2028-11-30pendingRobot foundation model hits 100M+ data hours — scaling-law inflectionHow: Figure or peer publishes that humanoid foundation model trained on >100M hours of robot or egocentric video — analog to GPT-3 inflectionSource: Project Go-Big internet-scale pretraining visionconf 45%
What if this resolves?
Click a button to clamp this prediction and run a Gibbs sample. Returns the predictions whose marginals shift most. ~30s per run; ideal for stress-testing "if X resolves, what else moves?"
Evidence chain
Network propagation neighbors
Top incoming (parents)
Edges that influence THIS node's belief
| Kind | Node | Their prob | P(c|s=T) | P(c|s=F) | Δ implied |
|---|---|---|---|---|---|
| prereq | S_HUMANOID_ENTERPRISE_2028 Humanoid R2: 100K+ enterprise by Nov 2028 | 50.0% | 0.500 | 0.050 | -0.080 |
| killer | TK08 Humanoid Capital Collapse (Figure/Apptronik Flop) | 22.0% | 0.050 | 0.500 | +0.046 |
Top outgoing (children)
Predictions THIS node influences
No outgoing edges.
Ticker exposure
Beneficiaries (24)
Adverse (5)
Prerequisites (6)
| Type | Pred | Title | Domain | Lag |
|---|---|---|---|---|
| prereq | S_HUMANOID_ENTERPRISE_2028 | Humanoid R2: 100K+ enterprise by Nov 2028 | humanoid_deployment | — |
| correlate | S_AGI_MID_2029 | AGI mid: Kurzweil 2029 path | agi_general_capability | — |
| correlate | S_AGI_FAST_2027 | AGI fast: drop-in remote worker by 2027-09 | agi_general_capability | — |
| correlate | S_AGI_SLOW_2031 | AGI slow: Schmidt/Hassabis 5-10 year path | agi_general_capability | — |
| correlate | S_AGI_WINTER_2036PLUS | AGI delayed: capability plateau or AI winter | agi_general_capability | — |
| killer | TK08 | Humanoid Capital Collapse (Figure/Apptronik Flop) | — | — |
Dependents (0)
| Type | Pred | Title | Domain | Lag |
|---|---|---|---|---|
| No dependents | ||||
Linked documents (10)
Raw metadata
{
"nia": false,
"url": "https://www.youtube.com/watch?v=S_fXhVT67Uw",
"mode": "PREDICTION",
"role": "Guest-CEO",
"context": "I hope this will lead to really um uh I think you'll see a lot of positive transfer emerge from a robot that's able to like generalized to a lot of things.",
"verbatim": "we've already shown positive transfer now with all this data. Like coming in the robot can journalize better with more information even like more knowledge is better... I hope this will lead to really um uh I think you'll see a lot of positive transfer emerge from a robot that's able to like generalized to a lot of things.",
"conv_cues": "I hope; I think",
"direction": "HAPPEN",
"timeframe": "future (unspecified)",
"conv_level": "MEDIUM",
"milestones": [
{
"kind": "quartile_checkpoint",
"label": "Q1 window check-in (25%)",
"status": "pending",
"weight": 0.05,
"ordinal": -8,
"source_id": null,
"expected_date": "2026-09-18",
"observed_date": null
},
{
"kind": "llm_pre_event",
"label": "Figure publishes positive-transfer benchmark showing diverse-data → broader task generalization",
"notes": "PARTIAL — Project Go-Big and Helix-02 already demonstrate cross-embodiment positive transfer (human video → robot navigation). Quantified benchmark across many tasks pending.",
"source": "Figure already demonstrated zero-shot human-to-robot transfer with Project Go-Big using egocentric human video",
"status": "partial",
"weight": 0.4,
"ordinal": -7,
"source_id": null,
"confidence": 0.85,
"source_url": "https://www.figure.ai/news/project-go-big",
"expected_date": "2026-11-14",
"observed_date": "2026-04-30",
"research_origin": "deep_research",
"expected_date_range": {
"to": "2027-06-30",
"from": "2026-04-01"
},
"measurement_criterion": "Figure / Helix research note quantifies that adding non-target-task data improves performance on held-out tasks (positive transfer coefficient > 1.0 across domains)"
},
{
"kind": "quartile_checkpoint",
"label": "Q2 window check-in (50%)",
"status": "pending",
"weight": 0.05,
"ordinal": -6,
"source_id": null,
"expected_date": "2027-02-06",
"observed_date": null
},
{
"kind": "llm_pre_event",
"label": "Helix 02 (or successor) demonstrates room-scale autonomy on novel tasks not in training distribution",
"source": "Helix 02 Full-Body Autonomy announcement; Adcock's 'end of C++' framing",
"status": "pending",
"weight": 0.4,
"ordinal": -5,
"source_id": null,
"confidence": 0.65,
"source_url": "https://www.humanoidsdaily.com/news/the-end-of-c-brett-adcock-on-helix-02-and-figure-s-path-to-room-scale-autonomy",
"expected_date": "2027-03-17",
"research_origin": "deep_research",
"expected_date_range": {
"to": "2027-12-31",
"from": "2026-06-01"
},
"measurement_criterion": "Figure releases technical report or video of Helix model executing 30+ unseen home/warehouse tasks zero-shot from natural language"
},
{
"kind": "llm_post_event",
"label": "Industry-wide acceptance of VLA (Vision-Language-Action) models as standard humanoid stack",
"source": "State of Robotics 2026 already calls VLA adoption the dominant 2026 trend",
"status": "pending",
"weight": 0.4,
"ordinal": -4,
"source_id": null,
"confidence": 0.8,
"source_url": "https://www.roboticscenter.ai/state-of-robotics-2026",
"expected_date": "2027-06-16",
"research_origin": "deep_research",
"expected_date_range": {
"to": "2028-06-30",
"from": "2026-06-01"
},
"measurement_criterion": "Tesla, 1X, Unitree, Agility all ship VLA-based stacks; transformer-based end-to-end policies become default robotics curriculum"
},
{
... (truncated)