How do we use this? · Background paper 2 · Argument 2 of 6
AI doesn’t need to be smarter. It needs a boss.
Background to the post of the same name, drawn from the working paper How Do We Use This?
Preprint. Reviewed by a panel of AI reviewers (Claude Opus 5.5) and revised 7 October 2026; not externally peer reviewed. Reproduced from the working paper as revised 7 October 2026.
In plain terms
The same models that pass hard benchmarks return nothing in most enterprise deployments, on MIT NANDA’s count: 95% of organizations get zero return from generative AI, a figure drawn from interviews and publicly criticized on method. The report attributes the divide, cautiously, to approach rather than model quality, and diagnoses a learning gap.
Our position: the missing piece is direction, not capability (reading the report’s “approach” as direction is ours). Direction is someone who loads what the agent needs before it starts, checks its work before it has consequences, and turns each correction into structure. No maker can supply that from where it sits. In the practice studied, the catch for three of five hand-picked misses in one working window began with the operator.
Published work finds the same models challenging users more in advice tasks and giving way more in debate when a reply is personalized. Our position, beyond what that work tested: the role is not fixed by the model; it is set by whoever sets up the work. That is a boss’s job.
About this paper. This is one of ten short background papers to a series of plain-English posts: six argument papers (background papers 1 to 6) and four primers (7, 8, 9, then 0). Every passage below is reproduced word for word from the working paper How Do We Use This? (Atkinson, 2026) as revised 7 October 2026, with that paper’s section, figure and table numbers. Passages are excerpts; a reference to anything not reproduced here resolves in the full paper. Tags such as “Our position” or “Conjecture” say what kind of support a claim has. Only the plain-terms summary, the box of terms, the post’s figure, and the section labels were written for this part. The full paper is linked from every post in the series.
Terms used here
Operator. The human who directs the agents and approves their consequential changes; here, the author.
Seat. The practice’s word for one running AI agent session, with its own context window and its own identifier.
Boot page, boot set. The boot page is the start-up page for the practice’s agent sessions; the boot set is the two rule-bearing documents every new session must load.
0-for-7. The boot page’s own record that a written rule was missed seven times on 12 July 2026, which it counts as one act rather than seven failures.
Stopping moment. The turn at which an agent states its answer as final without a further search or read.
Folding. A model giving up its position when challenged, whether or not the challenge is right.
The floor. A gap of one session between appearances of a tracked error class: it came back in the very next session. Recurrence, the program’s outcome measure, is how soon such a class comes back.
Detector, blind key. The detector is a local model that labels every operator turn as corrective or not. The blind key is a sample of turns the operator labeled by hand without seeing any model’s labels (195 graded rows); it measures how often the detector is right, not whether the agents’ answers were.
The counterweight. The paper’s proposed challenge to an agent at its stopping moment, not yet deployed. Its three clauses: (a) external to the agent’s own output; (b) firing on actual error, not on the agent’s confidence; (c) delivered at the stopping moment, at a rate above the ambient error rate.
H1 to H3. The paper’s hypotheses about the counterweight: H1, it is a projection of the operator’s judgment; H2, it is a control problem rather than a voting problem; H3, a reversal is not a correction.
From §4 · The proposition, and the evidence for it
4P2Direction, not capability
The operator’s proposition, stated as the practice’s working claim: “Your AI doesn’t need to be smarter. It needs to be led.”
A capable agent without direction behaves like a capable junior team that nobody was assigned to run. In this practice the operator directs the work, in person or through checks he built, approves every consequential change, and converts corrections into structure: a boundary that refuses, rather than a note that asks (§5.1). Whatever is failing when such agents go wrong, it is not intelligence; it is a leadership problem wearing a technology costume. The position would be overturned if the agents, unaided by the operator or his checks, caught most of their own premature stops. The counterweight of §8 is designed as a projection of that leadership, not a replacement for it (§8.4). This proposition is a reading of one practice and is offered as such.
4.1From the logsWho catches the misses
Table 3 lists five misses from one working window, reconstructed from the session transcript and version-control metadata. They were chosen to illustrate the failure’s shape and are not a sample. In each, the asserted state followed from the evidence the agent had gathered, and one further read would have falsified it.
| # | What the agent asserted | What the logs showed | Surfaced by |
|---|---|---|---|
| 1 | An empty handoff from a predecessor session indicated a defect. | The session was the first in a new line and had nothing to inherit. | Operator4 Sep, 10:08 |
| 2 | A named hook had closed the still-running session. | The record of the closure names what closed the session, and it was a different mechanism. | The agent, after the operator noted that an automated check had not fired4 Sep, 10:05 |
| 3 | Drafts in the operator’s voice could be written from five sample paragraphs. | A 57,400-word corpus of the operator’s writing was on the same disk. The agent had just read a note recording this same omission three days earlier, acknowledged it, and wrote two more drafts the same way. | Operator4 Sep, 14:51 |
| 4 | A rendered document was complete. | One page instead of thirteen, the wrong page size and typeface, at a plausible 23,782 bytes. | The agent’s own page count, within half a minute of rendering5 Sep, 05:26 |
| 5 | A pull request was open and blocking. | It had merged 49 minutes earlier; the agent had read a status post from an hour before as current. | An automated claim check at the end of the turn5 Sep, 05:28 |
Two of the five were surfaced by the operator outright, one by an automated check, one by the agent’s own check, and one by the agent only after the operator pointed at a check that had stayed silent. This is a tally of hand-picked cases, not a rate. The practice’s error log records only errors that were caught, so it can support a catcher share reported with its n and window, and never a catch rate. In these five, correction began with the operator in three and with a check he built in a fourth (case 5); case 4, the agent’s own page count, is the one unaided catch. Case 3 is the recurrence of §5.2 in miniature: the lesson was in the logs, and the agent read it, acknowledged it, and wrote two more drafts the same way.
4.2From recent workRole, not capability
Kelley and Riedl (2026) measure the epistemic effects of personalization across nine frontier models and find them role-dependent, grouping tasks as advisory (personal advice, moral judgment) or peer (debate, goal alignment). In open-ended advice, personalization made models more willing to challenge the user’s framing: the personalized response used more framing-accepting language than the generic one in 26.8% of paired comparisons. In debate it made them give way: highly personalized rebuttals raised the odds of a model abandoning its position about 2.4-fold over generic ones (logit β = 0.87), and accommodation tracked the persona’s inferred preference for validation (β = 0.57). Debate has no correct answer, so these shifts say nothing about truth; on factual questions (MMLU-Pro) personalization did not systematically amplify sycophancy. Where ground truth exists, a challenge such as “are you sure?” makes models abandon answers, correct ones included (Laban et al., 2023). A challenge can teach a model to fold as easily as to correct.
Mytsyk, Zhang and Krishnamurthy (2026) attack the same failure, a model giving way under a user’s pushback, from the training side rather than the harness side. Fine-tuning a 3-billion-parameter model (SmolLM3-3B) against a Bayesian truth serum reward, a scoring rule that pays for predicting what others will answer as well as for the answer itself, cut the answer-flip rate under user pressure from 23% to 4% and raised accuracy under that pressure from 80% to 93%, on the 100-question test split of a 1,000-item synthetic true-or-false set, in a single run; on four other small open models the effect was smaller and mostly not significant. Folding is therefore not a fixed property of a model, and something in it is trainable out. Their own conclusion is the half that matters more here: “Our results say nothing about correctness.” The reward tracks what the model believes, not what is true: by the authors’ account it pushes a response toward the answer the model’s own knowledge best supports, which “need not be the correct answer”. So a model can stop folding and stay wrong. That is H3 (§8.4: a reversal is not a correction), stated by authors who held the training lever this program does not. The agents studied here run on a commercial frontier family whose weights the practice does not hold, so the only surface available to it is the harness.
The same models challenged or gave way depending on the kind of task, and Kelley and Riedl note that a role can be set explicitly as well as inferred from a query’s framing. Our position, a step beyond what they tested: the role is not fixed by the model; it is set by whoever sets up the work, and that is a boss’s job. The gap this proposition names is one of direction, not capability, and a smarter model will not close it. It is also why the counterweight is built to speak as an advisor and never as a peer (§8.2).
The proposition also has survey-scale evidence from outside this practice. The MIT NANDA project’s industry study reports that despite $30–40 billion of enterprise investment, 95% of organizations are getting zero return from generative AI, with just 5% of integrated pilots extracting millions in value, against a review of 300+ public deployments, structured interviews at 52 organizations, and surveys of 153 senior leaders. The report calls its figures “directionally accurate based on individual interviews rather than official company reporting”, and the 95% was publicly criticized on methodological grounds (for example by Kevin Werbach of the Wharton School, as summarized in Futuriom, 26 August 2025). Its attribution is that the divide “does not seem to be driven by model quality or regulation, but seems to be determined by approach” (Challapally, Pease, Raskar & Chari, 2025). We read “approach” as direction, which puts this section’s claim at field scale: the same models that clear capability benchmarks return nothing where nothing directs them. Against it: the report’s own diagnosis is a learning gap, and it lists model output quality among the barriers to scaling.
From §2.3 · The September 2026 record
2.3The September 2026 context
Between the first of September 2026 and this revision (21 September 2026), the public discourse of the frontier laboratories shifted in a way that bears on this paper’s propositions, and the shift is recorded here with dates. On 2 September the chief executive of OpenAI described the coming generation of models as “sobering for everybody” and said that progress would from here be paced by alignment and safety work (Axios, 3 September 2026, interview at the G20 Innovation Ministerial, as carried by The Next Web). In a Fortune interview published on 12 September, the same executive said that “given everything happening with safety, right now would be an ill-advised moment to go public,” and ruled out an offering in 2026 (Fortune, 12 September 2026); both laboratories had been reported in May as preparing public offerings this year at estimated valuations of about $1 trillion each (Fortune, 26 May 2026). On 13 September, on CBS’s Sunday Morning, the chief executive of Anthropic said that “for too long the industry lied to people about the fact that this technology had risks,” called on the industry to slow capability development, and committed his company to permanent access for independent model evaluators (CBS News, 13 September 2026; the web article is stamped as updated 14 September, and CNBC, 13 September 2026, reports the same interview). Semiconductor equities fell on 14 September, which Reuters attributed to the accumulated statements while noting that the 10-year Treasury yield briefly passed 5% that day (Reuters, 14 September 2026, as carried by Yahoo Finance). Each of these is listed with its link under Press and public statements.
Two features of the laboratory leaders’ statements of 1 to 14 September 2026 matter here. First, each concerns what the models will be: more capable, more dangerous, sooner. None concerns how an organization is to use the models it already has, which is the question this practice exists to study, and which, we argue, no maker can answer from where it sits: how to use a model is a fact about the deploying organization’s work, its costs of error and its standards of correctness, none of which is visible from the laboratory. The reported incidents with autonomous AI agents in the same weeks (Fortune, 12 September 2026) bear on exactly that question. Second, no party to the September argument disputed the deployment evidence: the capability and risk claims moved markets while the reported failure rate of enterprise deployments (§4) stood unchallenged. We read the fortnight as corroboration, at the industry’s own scale, of the distinction between capability and direction that §4 draws. Against it: silence is weak evidence, and that failure rate is itself publicly disputed on method (§4). The fortnight is also an instance of §6’s subject: a maker’s public statement about its own unreleased model is self-report, authored by the measured party and not verifiable by its deployers before release. No result in this paper rests on any claim in this subsection.
From §5.1 · Onboarding: what is loaded before the work starts
What the record does support is a distinction the same boot page draws (the start-up page for the practice’s agent sessions, which records a rule going “0-for-7” on 12 July 2026), and it is sharper than that count: rules delivered one way drift; rules delivered another way fire.
“A step in a list is a rule, and rules went 0-for-7. Being first is a structure.”
The practice’s boot page, 16 July 2026
A rule an agent is supposed to go and read is a wish: it competes for attention with the work, and it loses. A rule injected into the context before the session’s first token is a property of the environment, and it does not have to win anything. The practice’s memory store is the second kind, loaded on every turn, and it fires. That is the boundary-versus-memo claim of §5 moved one level down: what matters is not whether a discipline is written but whether reading it is optional. This is our position, from one practice, and not a measured rate, because the measurement does not exist.
From §8.3 · Supervision: what the frameworks let a person approve
Two details from that survey of agent-framework defaults (September 2026) are worth stating on their own, because they come from the vendors rather than from us. Claude Code, whose agent loop the Claude Agent SDK embeds, ships a built-in completion condition in which a separate small model checks after each turn whether a stated goal has been met, and its documentation says of that evaluator that “it does not call tools, so it can only judge what Claude has already surfaced in the conversation” (Anthropic, 2026b). That is an accurate description of the limit this paper is about, published by the party with the most incentive to describe it favorably. The same page documents Stop hooks, an opt-in automated gate that runs a developer’s check each time the agent stops. The same vendor’s guidance on building agents says of having one model judge another that “this is generally not a very robust method” (Anthropic, 2025), while allowing that it can be worth its latency where any gain matters. Meanwhile the human-in-the-loop primitives in the OpenAI Agents SDK and the Claude Agent SDK gate tool calls and not completion claims (OpenAI’s output guardrails can check a final answer, if a developer writes them): a developer can require a person to approve an irreversible action before it is taken, and has nothing built in to require a person to confirm that the finished work was actually correct. CrewAI is the exception among the three current frameworks surveyed: a task can be set to have a human review the agent’s final answer, and like every other gate here that setting is off by default (CrewAI, 2026b). Outside that one opt-in, the approval surface exists for the act and not for the claim, which is the asymmetry the counterweight (this paper’s challenge mechanism, §8) is aimed at. This is not a criticism of those libraries, which are explicit about what they are; it is the baseline against which every mechanism in this paper should be read. The default case is not a weak check. It is no check until a developer adds one, and a stop the agent declares for itself.
From §8.4 · Feedback that becomes structure
H1, the counterweight is a projection of the operator’s judgment. The agents working under the harness generate a corpus of the operator’s corrections in the course of ordinary work; a nightly consolidation of the day’s corrections produces an updated projection of that judgment, and the projection, delivered into an agent’s context at the stopping moment, is the counterweight. The main path, delivery into the context window, needs neither weights nor logits. A secondary training path, steering a fixed model through a smaller tuned one (proxy-tuning, for example), needs the large model’s full next-token distribution at each step, which closed APIs generally do not expose; in practice it means open-weight models, and a review of model terms of use is a gate on it.
From §8.6 · The decisions that cannot be delegated
The counterweight is one instrument inside a set of decisions the practice cannot delegate, and the set deserves naming because, as we read the market, it is commoditizing everything around it. Over 2026 the infrastructure beneath working agents, the orchestration, retries and state handling, became a managed rental; this practice rents it and expects to keep renting it. What did not commoditize are the decisions that infrastructure exists to execute: whose knowledge becomes the standing instruction set; what result, fixed before work opens, ends an experiment; and what “done” means for a given piece of work. A practitioner essay contemporaneous with this program raised questions we read as these three, the unanswered residue of managed-agent platforms (Forstie, 2026, cited under Prior art); the wording is ours, not the essay’s. The practice’s answers are machinery this paper has already described: changes to the standing instruction set are proposed by the machine and merged or refused by the operator; experiments carry decision rules ratified before their data are seen, under which a null closes the program (§10.4); and the finish line is ruled in advance. The record is the load-bearing element in each: every one of those decisions is made from the practice’s own logs, which is why §7 treats ownership of the record as a first-order property rather than a storage preference. An execution history that lives on a platform’s dashboard, in the platform’s format, is testimony in the sense of §6, not a record the operator holds: retention, format and query access sit with the platform, and its summary of its own agents’ success is the vendor’s account of its own product.
From §13 · What would overturn each claim, and what transfers
| Rests on | Would be overturned by | Status | |
|---|---|---|---|
| P1 | Corrective signal flagged in ~48% of operator turns, roughly a third adjusted (§3.1); Mehta (2026) | A build that prices correctness at the stop and still terminates early; weakened if few corrections point to evidence the agent could have retrieved before it stopped | count measured; link inferred |
| P2 | The case series (§4.1; five hand-picked cases, not a rate); Kelley & Riedl (2026) | The same agents catching their own premature stops without operator direction | inferred |
| P3 | A tracked error class back in the very next session in 48 of 75 intervals (§5.2); the read-versus-injected mechanism, not a rate (§5.1); Wang & Huang (2026) | A remembered rule that holds across sessions where only a boundary now does | recurrence measured; link a conjecture |
| P4 | Load records in 75 of 579 sessions; the logging gap, about 19× adjusted (27× raw); four instrument failures (§6) | A self-report that tracks the logs it claims to summarize | counts measured; link inferred |
| P5 | The 8,221 operator turns of the six-month chat-lane log and the blind answer key (§2.2, §7) | A broad feedback dataset that links one owner’s corrections on consequential work to what followed, over months | conjecture |
What transfers, if anything does. Our position: three rules earned here rest on argument and on the work cited, not on this practice’s N of 1 alone. Rulebooks are theater; boundaries and incentives are not: a boundary that refuses and records its refusal outlasts a memo that asks an agent to remember (§5; from one practice, and the least tested of the three). A challenge must cite evidence the challenged agent can open, or it is an opinion wearing a citation (§8.2), because an unconstrained critic can invent its evidence (Kenton et al., 2026). And an outside checker must be calibrated against known-false cases before it is allowed to gate anything, because a checker built from the same class of model may simply agree with the claim it is shown (§9; Panickssery et al., 2024; Knight & Leveson, 1986). Conjecture: every team that clears its chat history lives inside its own version of the gap between the corrective signal its people supply and the corrections its system logs; that is the first thing a replication would count.
From §11 · Limits
External. One operator, one domain, one harness, one model family for the agents. Nothing here generalizes to other operators without replication, and the multiple-baseline extension (§10.2: staggered starts for the operator and three co-researchers) raises N to four within one organization, not to a population. The setting’s ecological validity is bought with exactly this cost.
Open questions. Readings of the working paper’s own records that cut against a position it takes, collected here rather than beside the positions they concern and left open; section numbers are the working paper’s.
- §5.1, loaded rules. On 3 and 4 September 2026 a notice printed as designed at the start of about forty consecutive turns was read and not acted on until a refusal was added on 4 September.
How to cite this part. Atkinson, B. (2026). AI doesn’t need to be smarter. It needs a boss. Background paper 2 to How do we use this? Working paper, Wolfberg LLC.
Disclosure. Drafting and literature synthesis were assisted by AI models (Claude, Anthropic) working under the author’s direction; the author is responsible for the content. The works in the References were read in full by AI reading agents working under the author’s direction, and quotations and numbers taken from them were checked against the saved full text. The prior-art works listed under Prior art are cited at the level of an established concept and its origin: each was verified for author, title, year and venue, but not read in full, and no numeric claim rests on any of them. Software documentation, source code and press accounts are listed under their own headings and were read at the linked pages. Every reference below carries a link, and every arXiv identifier and DOI was resolved against its registry, with title and first author matched, on 23 September 2026. A review by a panel of AI reviewers on 7 October 2026 found one quotation altered from its source (Mytsyk et al., 2026); it is corrected, and the checking step has been extended.
Competing interests. The author owns Wolfberg LLC, the practice studied.
Data availability. The practice’s logs contain client work and personal records. They are private and are not offered for sale or sharing. The measures are described in enough detail to be reimplemented, and figures from the practice are reported as of the dates given. What is available is the design: the clauses of §8.1, the evidence contract and admission checks of §8.5, and the decision rules of §10.4 are stated in this working paper fully enough to be rebuilt without access to the logs; the background papers reproduce only parts of them.
Corrections to the full working paper. The notes below record corrections to the working paper as a whole; some concern sections not reproduced in this part.
Corrections, 23 September 2026. No figure and no finding changed. The paper was retitled; earlier revisions were titled The Stop Problem: Defensible Is Not Correct, and the stop problem remains this paper’s name for the failure it studies. The Anthropic interview in §2.3 aired on 13 September, not over a weekend of 13 and 14 September; the web article is stamped 14 September. The essay listed under Prior art is by Ryan Forstie; an earlier revision gave the initial K. The completion evaluator in §8.3 is documented as a Claude Code feature, whose agent loop the Claude Agent SDK embeds; an earlier revision attributed it to the SDK directly. Two works in the References, Graves (2016) and Liu (2026), were listed without being cited in the text; each is now cited where it bears (§2.1, §8.3). Links were added to every press, documentation and prior-art source. A duplicated section number in §10 was corrected.
Corrections, 25 September 2026. No figure changed. §8.3 said that the human-in-the-loop primitives across the three frameworks surveyed gave a developer nothing to require a person to confirm finished work. That holds for the OpenAI Agents SDK and the Claude Agent SDK. It does not hold for CrewAI, whose task documentation, already cited as CrewAI (2026b), offers an opt-in setting for a human to review the agent’s final answer; §8.3 now says so.
Corrections, 7 October 2026. After review by a panel of AI reviewers: a quotation in §4.2 from Mytsyk, Zhang and Krishnamurthy (2026) carried words its source does not contain (“or truthfulness, only about sycophancy”) and now reads as the source does, “Our results say nothing about correctness”; the point it supports is unchanged. The §8.3 design rule is no longer stated as a law: an instrument the measured party can operate tends to rot; one it cannot alter is harder to game, though not immune. Zhang et al. (2026) is now read as a gain from training, not from moving a check outside the agent (§8.3, Figures 10 and 11). Laban et al. (2023) replaces Kelley and Riedl (2026), whose tasks have no ground truth, in Table 5 and H3. The 48% is now the share of operator turns the detector flagged, roughly a third adjusted; the 27× logging gap is about 19× adjusted and partly definitional. Table 9 separates measured counts from interpreted links, §13 offers its design rules as hypotheses, and the Disclosure says who read the cited works. A second pass the same day restored, in the author’s wording, positions that revision had softened for reasons other than fact, and labelled each for status (our position, conjecture, from one practice, measured) instead of hedging it; Table 9 now carries a Status column and §13 states its design rules as our position. Every factual correction above stands.
Works cited in this part (keys, including letter suffixes such as 2026b, follow the full paper’s reference list)
References
- Challapally, A., Pease, C., Raskar, R., & Chari, P. (2025). The GenAI Divide: State of AI in Business 2025. MIT NANDA project, July 2025. Project page: nanda.media.mit.edu; the report as read: mlq.ai/media/quarterly_decks/v0.1_State_of_AI_in_Business_2025_Report.pdf
- Kelley, S. W., & Riedl, C. (2026). Personalization increases affective alignment but has role-dependent effects on epistemic independence in LLMs. arXiv:2603.00024. arxiv.org/abs/2603.00024
- Kenton, Z., Janzer, L., Greig, R., Teh, T. H., Tyshchuk, K., Brown-Cohen, J., Edwards, H., Rajamanoharan, S., Siegel, N. Y., Jaques, N., et al. (2026). Debate training reduces reward hacking in RLAIF. arXiv:2608.17776. arxiv.org/abs/2608.17776
- Laban, P., Murakhovs’ka, L., Xiong, C., & Wu, C.-S. (2023). Are you sure? Challenging LLMs leads to performance drops in the FlipFlop experiment. arXiv:2311.08596. arxiv.org/abs/2311.08596
- Mehta, A. (2026). When agents commit too soon: Diagnosing premature commitment in LLM agents. Snowflake AI Research. arXiv:2606.22936. arxiv.org/abs/2606.22936
- Mytsyk, S., Zhang, Y., & Krishnamurthy, V. (2026). Mitigating LLM sycophancy with RL-based fine-tuning: Bayesian truth serum approach. arXiv:2608.25267. arxiv.org/abs/2608.25267
- Panickssery, A., Bowman, S. R., & Feng, S. (2024). LLM evaluators recognize and favor their own generations. arXiv:2404.13076. arxiv.org/abs/2404.13076
- Wang, J., & Huang, J. (2026). Reward hacking as equilibrium under finite evaluation. arXiv:2603.28063. arxiv.org/abs/2603.28063
Prior art (cited at concept level; verified for author, title, year and venue, not read in full)
- Knight, J. C., & Leveson, N. G. (1986). An experimental evaluation of the assumption of independence in multiversion programming. IEEE Transactions on Software Engineering, SE-12(1), 96–109. doi.org/10.1109/TSE.1986.6312924
- Forstie, R. (2026). The part of the agent stack nobody wants to build. LinkedIn, 25 August 2026. linkedin.com/pulse/part-agent-stack-nobody-wants-build-ryan-forstie-ihyqc Cited for its three closing questions on managed-agent platforms, first read on 1 September 2026; author, title, date and the three questions re-verified at the source on 23 September 2026. No figure from it is cited anywhere in this paper.
Software and documentation (read at the linked pages, September 2026; current as of then, not permanent)
- Anthropic. (2025, September 29). Building agents with the Claude Agent SDK. claude.com/blog/building-agents-with-the-claude-agent-sdk
- Anthropic. (2026b). Keep Claude working toward a goal. Claude Code documentation. code.claude.com/docs/en/goal
- CrewAI. (2026b). Tasks. CrewAI documentation. docs.crewai.com/en/concepts/tasks
Press and public statements (§2.3; read at the linked pages, or at the named carrier where the original is paywalled)
- Axios. (2026, September 3). Sam Altman’s sobering siren. Interview at the G20 Innovation Ministerial. axios.com/2026/09/03/axios-interview-sam-altmans-sobering-siren; paywalled, read as carried by The Next Web: thenextweb.com/news/sam-altman-axios-idea-guy-sobering-models
- CBS News. (2026, September 13). Anthropic CEO Dario Amodei: “For too long the industry lied” about AI risks. Sunday Morning; web article updated 14 September. cbsnews.com/news/anthropic-ceo-dario-amodei-on-ai-risks/
- CNBC. (2026, September 13). Anthropic’s Amodei says China presents “toughest dilemma” for his proposed AI slowdown. Cited for the broadcast date. cnbc.com/2026/09/13/china-dilemma-ai-slowdown-anthropic.html
- Fortune. (2026, May 26). On the two laboratories walking back earlier job-loss forecasts ahead of public offerings; cited for the reported valuations. fortune.com/2026/05/26/sam-altman-dario-amodei-walking-back-ai-jobs-apocalypse-prophecies-ipo/
- Fortune. (2026, September 12). Interview with the chief executive of OpenAI on safety and the timing of a public offering. fortune.com/2026/09/12/sam-altman-interview-ai-doomsday-safety-models-control-ipo-2027/
- Futuriom. (2025, August 26). Why we don’t believe MIT NANDA’s weird AI study. By R. Scott Raynovich; summarizes Kevin Werbach’s LinkedIn critique of the 95% figure. futuriom.com/articles/news/why-we-dont-believe-mit-nandas-werid-ai-study/2025/08
- Reuters. (2026, September 14). Wall Street ends down, calls for AI slowdown pummel chipmakers. As carried by Yahoo Finance. finance.yahoo.com/technology/ai/articles/ai-warnings-knock-nasdaq-futures-092329455.html