← Cockpit
233_021predictionAIAGI

AI learning will improve via closed-loop reinforcement learning cycle making results keep increasing.

Predictor: Joe Liemandt · ep#233 "This $40M AI Company Is Using AI Tutors to Teach 2 Hours/Day | #233" · source

Prior probability
45.0%
Current probability
38.7%
evolves via intake + LBP
Conviction
4/5
Signal quality
C
Resolution
pending
Window
2028-06-01 – 2028-06-30
Edges in / out
11 / 5
Tickers exposed
21

Prediction text

AI learning will improve via closed-loop reinforcement learning cycle making results keep increasing. | what we have now and why our learning keeps increasing is we have the we're the only ones in the world with a closedloop data cycle. >> It's reinfor it's reinforcement learning for kids.

Watch events: ARC-AGI-2 scores; Frontier Math Tier 4 benchmark; SWE-bench Verified; Humanity's Last Exam

Verbatim quote

From episode "This $40M AI Company Is Using AI Tutors to Teach 2 Hours/Day | #233"
what we have now and why our learning keeps increasing is we have the we're the only ones in the world with a closedloop data cycle. >> It's reinfor it's reinforcement learning for kids.

Predictor: Joe Liemandt

κ + Brier as of 2026-05-22
κ (discount)
0.583
Brier
0.0064
excellent
Hits / Misses
1 / 0
of 1 resolved
Hit rate
100.0%
Calibration plot (stated vs observed)

Evidence about this node from Joe Liemandt 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

Not linked

This node isn't linked to a reference class. The Bayesian update applies without outside-view blending.

Probability over time

3 prob_history rows
0%25%50%75%100%prior 45%2026-04-302026-04-302026-05-03
intake v2milestone miss sweeplbp propagationreference class assignedlegacy v1prior_prob (analyst seed)current = 38.7%

Milestone chain

Pre-event signals (upstream prereqs + window checkpoints) → resolution event → downstream cascades. Status/dates update from linked nodes; re-derive nightly via scripts/ops/derive_milestones.py.
Leading chain: 5 fired ✓ · 6 pending
  1. 2026-03-01hitAlpha students rank top 1% nationally via AI-tutor closed-loop system
    How: Independent third-party standardized test results show Alpha cohort scoring >=99th percentile vs national norms (validated, repeated)
    Source: Peter Attia podcast #366 with Joe Liemandt; Alpha School public materialsconf 85%
    Notes: Empirical validation of closed-loop RL learning claim — multi-year top-1% performance is a strong signal.
  2. 2026-01-01 → 2027-06-30pendingAlpha School scales 'Time Back' AI tutor platform to additional schools
    How: Alpha School / Trilogy publishes deployment of closed-loop AI tutor at >=10 partner schools beyond original Austin location
    Source: Alpha School: Joe Liemandt's Vision for Alpha School (2026)conf 70%
  3. 2026-06-01 → 2027-12-31pendingLiemandt's $1B commitment funds sub-$1,000 device deployment to scale
    How: Public reporting confirms shipment of >=10,000 sub-$1,000 tablets running closed-loop AI tutor with deployment data
    Source: Mike Maples / Floodgate notes; Alpha School roadmapconf 40%
  4. 2026-09-01 → 2027-12-31pendingMainstream LLM lab releases closed-loop RL tutoring product
    How: OpenAI, Anthropic, Google, or xAI launches consumer/education-targeted closed-loop tutor with RL feedback (analog to Alpha's approach)
    Source: Vendor product announcementconf 55%
  5. 2027-01-01 → 2028-06-30pendingClosed-loop AI tutor matches/exceeds human instructor on rigorous benchmarks
    How: Peer-reviewed RCT shows closed-loop AI tutor delivers >=2x learning velocity vs traditional classroom on standardized benchmark (e.g., NAEP, TIMSS-aligned)
    Source: Education research journal (e.g., Educational Researcher, Cognition and Instruction)conf 45%
  6. 2036-09-06pendingMoon base will exist in 10 years

What if this resolves?

Clamp this prediction TRUE or FALSE and run a counterfactual Gibbs sample. Surfaces the predictions whose marginals shift most under that assumption.
(live posterior: 39%)

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

Every probability update with full Bayesian provenance — chronological, latest first
LBP2026-05-03T02:00:01Z38.7%-1.2pp
Network propagation: 39.9% → 38.7%
6-iter LBP, residual 0.00677 · damping 0.5, w_intrinsic 0.5 · method lbp_v3 · run 1a683ac9
LBP2026-04-30T16:39:51Z39.9%-1.5pp
Network propagation: 41.4% → 39.9%
5-iter LBP, residual 0.00825 · damping 0.5, w_intrinsic 0.5 · method lbp_v2 · run 0c8a4ea3
LBP2026-04-30T02:18:57Z41.4%-3.6pp
Network propagation: 45.0% → 41.4%
5-iter LBP, residual 0.00825 · damping 0.5, w_intrinsic 0.5 · method lbp_v1 · run 592311ef

Network propagation neighbors

Top edges sorted by latest LBP cross-impact
All propagation →

Top incoming (parents)

Edges that influence THIS node's belief

KindNodeTheir probP(c|s=T)P(c|s=F)Δ implied
prereqCMQ_001
By 2026, AI will reach 'intern-level' capability — millions Sam Altman
44.8%0.4500.050-0.133
prereq248_040
Pausing AI will fail and only accelerate race dynamics.Alex Wissner-Gross
53.0%0.4500.050-0.128
prereq232_014
Recursive self-improvement is already here, not 12 months awAlex Wissner-Gross
70.2%0.4500.050-0.056
prereq235_038
David Sinclair begins partial epigenetic reprogramming trialPeter Diamandis
74.0%0.4500.050-0.044
prereq238_009
Recursive self-improvement is already happening now (no longAlex Wissner-Gross
78.1%0.4500.050-0.028

Top outgoing (children)

Predictions THIS node influences

KindNodeTheir probP(c|s=T)P(c|s=F)Δ implied
prereq239_001
Global economy will be 10x its current size in 10 yearsElon Musk
37.7%0.6000.050-0.121
prereq235_030
Ray Kurzweil predicts Longevity Escape Velocity (LEV) by 203Ray Kurzweil
39.2%0.7500.050-0.080
prereqCMQ_003
By 2030, AI models will surpass peak human expert levels acrSam Altman
22.8%0.3500.050-0.065
prereq239_008
Moon base will exist in 10 yearsElon Musk
28.8%0.5500.050-0.050
prereqSEM_034
True artificial general intelligence will be achieved betweeDemis Hassabis
28.7%0.5500.050-0.049

Ticker exposure

21 ticker(s) linked

Beneficiaries (14)

SOUNNVDAGTLBAIBBAITCEHYAMZNBABAGOOGLIBMMETAMSFTORCLSHOP

Adverse (7)

ACNCTSHFRSHCHGGIBMINFYPEGA

Prerequisites (11)

Predictions that must hit first
TypePredTitleDomainLag
prereq235_038David Sinclair begins partial epigenetic reprogramming trials with Life Biosciences in March 2026.Biotech/Longevity
prereq232_014Recursive self-improvement is already here, not 12 months away.AI
prereq248_040Pausing AI will fail and only accelerate race dynamics.AI
prereq238_009Recursive self-improvement is already happening now (no longer three years out)AI
prereqCMQ_001By 2026, AI will reach 'intern-level' capability — millions of virtual interns performing supervised, economically useful tasks.AI
correlateS_AGI_MID_2029AGI mid: Kurzweil 2029 pathagi_general_capability
correlateS_AGI_FAST_2027AGI fast: drop-in remote worker by 2027-09agi_general_capability
correlateS_AGI_SLOW_2031AGI slow: Schmidt/Hassabis 5-10 year pathagi_general_capability
correlateS_AGI_WINTER_2036PLUSAGI delayed: capability plateau or AI winteragi_general_capability
killerTK01AGI Capability Plateau (2026-27 Training Stall)
killerTK03AI Regulatory Moratorium (EU/US Capability Freeze)

Dependents (5)

Predictions enabled by this
TypePredTitleDomainLag
prereq235_030Ray Kurzweil predicts Longevity Escape Velocity (LEV) by 2033.Biotech/Longevity
prereq239_001Global economy will be 10x its current size in 10 yearsMacro/Economy
prereqSEM_034True artificial general intelligence will be achieved between 2032 and 2042 — 'first we solve AI, then use AI to solve everything else'.AI/AGI
prereq239_008Moon base will exist in 10 yearsSpace
prereqCMQ_003By 2030, AI models will surpass peak human expert levels across virtually all cognitive domains — onset of true superintelligence.AI

Linked documents (10)

Auto-generated by cosine similarity from Polymarket / Manifold / EDGAR / GDELT

Raw metadata

From Thesis_Timeline_v1.0_FINAL workbook
{
  "nia": false,
  "url": "https://www.youtube.com/watch?v=X94eBT-VZnc",
  "mode": "THESIS",
  "role": "Guest-CEO",
  "context": "taking that last part that you were talking about the data what we have now and why our learning keeps increasing is we have the we're the only ones in the world with a closedloop data cycle. It's reinforcement learning for kids. It is. And so our our learning science team comes up with an idea and they're like, \"Okay, let's implement this new idea and see.\"",
  "to_year": 2030,
  "verbatim": "what we have now and why our learning keeps increasing is we have the we're the only ones in the world with a closedloop data cycle. >> It's reinfor it's reinforcement learning for kids.",
  "conv_cues": "learning keeps increasing",
  "direction": "UP",
  "from_year": 2026,
  "timeframe": "ongoing",
  "conv_level": "HIGH",
  "milestones": [
    {
      "kind": "llm_pre_event",
      "label": "Alpha students rank top 1% nationally via AI-tutor closed-loop system",
      "notes": "Empirical validation of closed-loop RL learning claim — multi-year top-1% performance is a strong signal.",
      "source": "Peter Attia podcast #366 with Joe Liemandt; Alpha School public materials",
      "status": "hit",
      "weight": 0.4,
      "ordinal": -11,
      "source_id": null,
      "confidence": 0.85,
      "source_url": "https://peterattiamd.com/joeliemandt/",
      "expected_date": "2026-03-01",
      "observed_date": "2026-03-01",
      "research_origin": "deep_research",
      "measurement_criterion": "Independent third-party standardized test results show Alpha cohort scoring >=99th percentile vs national norms (validated, repeated)"
    },
    {
      "kind": "prereq",
      "label": "Recursive self-improvement is already happening now (no longer three years out)",
      "status": "hit",
      "weight": 0.5,
      "ordinal": -10,
      "source_id": "238_009",
      "expected_date": "2026-04-29",
      "observed_date": "2026-04-29"
    },
    {
      "kind": "prereq",
      "label": "David Sinclair begins partial epigenetic reprogramming trials with Life Biosciences in March 2026.",
      "status": "hit",
      "weight": 0.5,
      "ordinal": -9,
      "source_id": "235_038",
      "expected_date": "2026-04-29",
      "observed_date": "2026-04-29"
    },
    {
      "kind": "prereq",
      "label": "Pausing AI will fail and only accelerate race dynamics.",
      "status": "hit",
      "weight": 0.5,
      "ordinal": -8,
      "source_id": "248_040",
      "expected_date": "2026-04-29",
      "observed_date": "2026-04-29"
    },
    {
      "kind": "prereq",
      "label": "Recursive self-improvement is already here, not 12 months away.",
      "status": "hit",
      "weight": 0.5,
      "ordinal": -7,
      "source_id": "232_014",
      "expected_date": "2026-04-29",
      "observed_date": "2026-04-29"
    },
    {
      "kind": "prereq",
      "label": "By 2026, AI will reach 'intern-level' capability — millions of virtual interns performing supervised, economically useful tasks.",
      "status": "pending",
      "weight": 0.5,
      "ordinal": -6,
      "source_id": "CMQ_001",
      "expected_date": "2026-06-26",
      "observed_date": null
    },
    {
      "kind": "llm_pre_event",
      "label": "Alpha School scales 'Time Back' AI tutor platform to additional schools",
      "source": "Alpha School: Joe Liemandt's Vision for Alpha School (2026)",
      "status": "pending",
      "weight": 0.4,
      "ordinal": -5,
      "source_id": null,
      "confidence": 0.7,
      "source_url": "https://alpha.school/the-future-of-education-joe-liemandts-vision-for-alpha-school/",
      "expected_date": "2026-09-30",
      "research_origin": "deep_research",
      "expected_date_range": {
        "to": "2027-06-30",
        "from": "2026-01-01"
      },
      "measurement_criterion": "Alpha School / Trilogy publishes deployment of closed-loop AI tutor at >=10 partner schools beyond original Austin location"
    },
    {
... (truncated)