Site icon Decode the future

Prospect Theory Explained: Kahneman & Tversky’s Model, With Examples

Last updated: October 2026

Prospect theory is Daniel Kahneman and Amos Tversky’s 1979 model of how people actually choose under risk. We judge outcomes as gains or losses from a reference point, feel losses about twice as strongly as gains, gamble to avoid sure losses, and overweight small probabilities — which is why people buy both insurance and lottery tickets.

Behavioral Economics Kahneman & Tversky Cumulative Prospect Theory Decision Under Risk

Prospect theory is the most influential descriptive model of decision-making under risk. It replaced the idea of a perfectly rational chooser with four measurable regularities — and it is the reason behavioral economics exists as a field. This guide covers the full model: the value function, probability weighting, the reflection effect, framing, the 1992 cumulative prospect theory upgrade, how it differs from expected utility theory, and whether it still replicates in 2026.

Prospect theory vs loss aversion — which page do you need?

Prospect theory is the whole model. Loss aversion is one of its four parts. If you only need loss aversion — the coefficient λ, everyday examples and the 2024–2025 meta-analyses on whether λ ≈ 2 holds up — read loss aversion explained: definition, examples and the λ debate.

What is prospect theory?

Prospect theory is a behavioral economics model that describes how people actually make decisions when outcomes are uncertain, as opposed to how classical economics says they should decide. Before 1979 the dominant framework was expected utility theory, which treats people as rational agents who weigh final wealth states by their objective probabilities. Experiments such as the Allais paradox had already shown systematic violations. Kahneman and Tversky’s paper “Prospect Theory: An Analysis of Decision under Risk” (Econometrica, 1979) turned those violations into a coherent alternative model.

The paper is one of the most cited in economics. Kahneman received the 2002 Nobel Memorial Prize in Economic Sciences largely for this work. Tversky died in 1996, and the Nobel is not awarded posthumously. Kahneman died on 27 March 2024. The theory’s name is deliberately plain: a “prospect” is simply a gamble — a list of outcomes with their probabilities.

Kahneman and Tversky’s prospect theory: the four building blocks

Building block What it says Everyday consequence
1. Reference dependence Outcomes are coded as gains or losses relative to a reference point, not as final wealth A $5,000 bonus feels like a loss if you expected $10,000
2. Loss aversion Losses weigh more than equal gains (λ ≈ 2–2.25) Rejecting a 50/50 bet to win $150 or lose $100
3. Diminishing sensitivity Each extra dollar matters less, for gains and for losses Risk-averse for gains, risk-seeking for losses (the reflection effect)
4. Probability weighting Small probabilities are overweighted, moderate and large ones underweighted Buying lottery tickets and insurance

Prospect theory also describes two phases of choice. In the editing phase people simplify the options: they set a reference point, code outcomes as gains or losses and round probabilities. This is where framing effects enter. In the evaluation phase they score each edited prospect using the value function and the probability weights, and pick the highest. Neither step is a conscious calculation — it is the intuitive “System 1” Kahneman later described in Thinking, Fast and Slow (2011).

How does the prospect theory value function work?

The value function is the mathematical core of prospect theory. It replaces the utility-of-wealth curve from classical economics with an S-shaped curve defined over gains and losses.

Reference dependence

People don’t evaluate outcomes in absolute terms (“I have $50,000”). They evaluate them as gains or losses relative to a reference point — usually the status quo, a purchase price or an expectation. If your portfolio was worth $100,000 last month and is now $95,000, you feel a $5,000 loss, even if $95,000 is more than you had a year ago.

In CFD trading the reference point is usually your entry price. I’ve experienced this first-hand on Plus500: a gold position opened at $2,340 that drops to $2,320 feels like a loss even if the thesis is intact. The reference point anchors the emotional response regardless of the analysis.

Loss aversion (λ)

The curve is steeper for losses than for gains. In Tversky and Kahneman’s 1992 estimates, the loss side is λ = 2.25 times as steep, so a fair coin flip on $100 feels like a bad deal. Loss aversion is the best-known part of the model — and the most debated. A 2024 Journal of Economic Literature meta-analysis puts the average near 1.96, while some newer designs find much smaller values. We cover that evidence, and examples from pricing, insurance and housing, in our dedicated guide to loss aversion and the loss aversion coefficient. One of its best-known consequences is the endowment effect.

Diminishing sensitivity

The value function is concave for gains (risk-averse) and convex for losses (risk-seeking). The difference between gaining $100 and $200 feels larger than the difference between gaining $1,100 and $1,200. The same applies to losses, which is why people who are already deep in a losing position often double down (“I’m already $5,000 down — what’s another $1,000?”).

Prospect Theory Value Function — S-Shaped Curve The prospect theory value function showing an S-shaped curve that is concave for gains (above reference point) and steeper convex for losses (below reference point), illustrating loss aversion with λ ≈ 2.0–2.5. The kink at the origin represents the reference point. Prospect Theory Value Function DecodeTheFuture.org prospect theory, value function, loss aversion, reference point S-shaped value function showing asymmetric response to gains vs losses, with loss aversion coefficient λ ≈ 2.0–2.5. Diagram image/svg+xml en © DecodeTheFuture.org Value (Psychological) Outcome Gains (concave, risk-averse) Losses (convex, risk-seeking) Reference Point Loss aversion: λ ≈ 2.0–2.5 Losses slope is ~2× steeper than gains slope Steeper Flatter

Tversky and Kahneman (1992) gave the value function a simple power form and estimated median parameters from their experiments:

v(x) = xα for x ≥ 0   |   v(x) = −λ · (−x)β for x < 0   with α = β = 0.88, λ = 2.25

What is the reflection effect in prospect theory?

The reflection effect is the finding that risk preferences flip when the same choice is moved from gains to losses. In the original 1979 paper, 80% of respondents preferred a sure $3,000 to an 80% chance of $4,000 — risk aversion. When the signs were reversed, 92% preferred an 80% chance of losing $4,000 to a sure loss of $3,000 — risk seeking.

Kahneman & Tversky (1979) Option A (sure) Option B (gamble) Majority choice
Gains (Problem 3) $3,000 for sure 80% chance of $4,000 A — 80% (risk-averse)
Losses (Problem 3′) −$3,000 for sure 80% chance of −$4,000 B — 92% (risk-seeking)

The gamble has the higher expected value in the gains problem ($3,200 vs $3,000) and the worse one in the losses problem (−$3,200 vs −$3,000). People get it “wrong” in both directions. That follows directly from the S-shaped value function: concave above the reference point, convex below it. The reflection effect is why traders lock in small profits but let losses run — the disposition effect.

What is probability weighting and why does it matter?

The second core mechanism is the probability weighting function. People don’t use objective probabilities when they choose. They transform them in a systematic, inverse-S pattern: small probabilities are overweighted and moderate-to-large ones are underweighted. Certainty gets extra weight — the certainty effect. Tversky and Kahneman (1992) used this one-parameter form:

w(p) = pγ / (pγ + (1 − p)γ)1/γ   with γ = 0.61 for gains, δ = 0.69 for losses

With the gains parameter, a 1% chance feels like about 5.5%, a 50% chance like about 42%, and a 99% chance like about 91%. This single mechanism explains two behaviors that look contradictory in the same person: buying lottery tickets (overweighting a tiny chance of a big win) and buying insurance (overweighting a tiny chance of a big loss). Under expected utility theory the two are hard to reconcile. Under prospect theory they come from the same distortion.

The four-fold pattern of risk attitudes

Combine the value function with probability weighting and you get prospect theory’s best-known prediction. People are risk-averse for high-probability gains (take a sure $450 over a 50% shot at $1,000) but risk-seeking for low-probability gains (buy lottery tickets). The pattern reverses for losses: risk-seeking for high-probability losses (gamble rather than accept a sure loss) and risk-averse for low-probability losses (insure against rare disasters).

High probability Low probability
Gains Risk-averse → “Take the sure thing” (sell winning stocks early) Risk-seeking → “Go for the jackpot” (lottery tickets, meme stocks)
Losses Risk-seeking → “Gamble to avoid the loss” (hold losing positions) Risk-averse → “Protect at all costs” (insurance, extended warranties)

Framing: how the same choice gets a different answer

Because outcomes are coded relative to a reference point, how a choice is described can move it from the gains frame to the losses frame — and flip the decision. Tversky and Kahneman’s “Asian disease problem” (Science, 1981) is the classic demonstration. Facing an outbreak expected to kill 600 people, 72% chose a programme that would “save 200 people” for sure over a one-in-three chance to save all 600. When the identical options were described as “400 people will die” for sure versus a one-in-three chance that nobody dies, 78% chose the gamble.

The same effect appears in medicine. McNeil, Pauker, Sox and Tversky (New England Journal of Medicine, 1982) showed patients, students and physicians the same lung-cancer outcome data for surgery versus radiation therapy. Surgery became more attractive when results were framed as chances of surviving rather than chances of dying — even among physicians.

Framing is also how price anchors work: the first number you see becomes the reference point that defines what counts as a gain or a loss. See the anchoring effect and mental accounting, Richard Thaler’s extension of prospect theory to everyday budgeting.

What is cumulative prospect theory (CPT)?

Cumulative prospect theory is the 1992 upgrade of the model, published by Tversky and Kahneman as “Advances in Prospect Theory: Cumulative Representation of Uncertainty” (Journal of Risk and Uncertainty). It is the version economists use today when they estimate parameters or build models.

Why was it needed? The 1979 version weighted each outcome’s probability on its own. With a non-linear weighting function, that can make a prospect look better even when another prospect is better in every state — a violation of first-order stochastic dominance. The 1979 paper patched this with editing rules. CPT fixes it properly, building on John Quiggin’s rank-dependent utility (1982). Weights are applied to cumulative probabilities of outcomes ranked from best to worst, separately for gains and losses.

Feature Original prospect theory (1979) Cumulative prospect theory (1992)
Probability weighting Applied to each outcome’s probability separately Applied to cumulative (rank-ordered) probabilities
Stochastic dominance Can be violated; relies on editing rules Always respected
Number of outcomes Designed for simple prospects with up to two non-zero outcomes Any number of outcomes, risk (known probabilities) and uncertainty
Gains vs losses One weighting function Separate weighting functions (γ = 0.61, δ = 0.69)
Parameters Qualitative shape Estimated: α = β = 0.88, λ = 2.25, γ = 0.61, δ = 0.69

In practice, when a paper reports “λ = 2.25” or “γ = 0.61”, it is using CPT. CPT is also the version used in asset pricing. Barberis, Jin and Wang (Journal of Finance, 2021) built a CPT-based model of investor preferences and found it helps explain a majority of 23 well-known stock-market anomalies. It works best for momentum, volatility, distress and profitability effects, and poorly for the value anomaly.

How does prospect theory differ from expected utility theory?

Dimension Expected Utility Theory Prospect Theory
What is evaluated Final wealth states Changes from a reference point (gains/losses)
Risk attitude Consistent (usually risk-averse everywhere) Risk-averse for gains, risk-seeking for losses (reflection effect)
Probability handling Objective probabilities, used linearly Decision weights: small probabilities overweighted, large ones underweighted
Loss vs. gain Symmetric treatment Loss aversion (λ ≈ 2–2.25)
Description of options Irrelevant (invariance) Matters (framing effects)
Nature Normative (how you should decide) Descriptive (how you actually decide)
Explains insurance + lottery Hard (needs an unusual utility function) Yes (probability weighting explains both)

The key shift: expected utility theory asks “Which option gives the best final outcome on average?” Prospect theory asks “Which option feels best relative to where I am now?” That shift explains decades of anomalies that classical economics struggled with.

7 real-world applications of prospect theory

1. The disposition effect in investing

The disposition effect, named by Shefrin and Statman (1985), is the most documented application in markets: investors sell winners too early and hold losers too long. Above the entry price you are in the concave gains region, so locking in a sure gain is attractive. Below it you are in the convex loss region, so holding on and hoping beats accepting a sure loss. Liu et al. (PLoS ONE, 2014) analysed 28.5 million trades by about 81,000 traders on a social-trading platform and found patterns consistent with both the reflection effect and loss aversion. Full guide: disposition effect explained (Polish: efekt dyspozycji).

2. Insurance and lotteries

The same person who buys insurance (paying a small sure loss to avoid a rare large loss) also buys lottery tickets (paying a small sure loss for a rare large gain). Probability weighting explains both: rare events get far more weight than their odds justify.

3. The equity premium puzzle

Stocks have historically returned about 6 percentage points a year more than safe bonds — far more than standard models of risk can justify. Benartzi and Thaler (1995) showed that loss-averse investors who evaluate their portfolios about once a year would demand roughly this premium. Frequent evaluation means seeing more paper losses, which weigh about twice as much as gains.

4. Status quo bias and defaults

Switching from the current state involves potential losses (which loom large) and potential gains (which feel smaller), so people stick with defaults even when alternatives are better (Samuelson & Zeckhauser, 1988). Policy designers use this deliberately: automatic enrolment and Save More Tomorrow make saving the default and tie contribution increases to pay rises, so take-home pay never visibly falls. See also nudge theory.

5. Sunk costs

After investing $10,000 in a failing project, you are in the domain of losses. Quitting means accepting a definite loss, while continuing offers a small chance of recovery. Diminishing sensitivity means the extra potential loss doesn’t feel proportionally worse, so people throw good money after bad.

6. Salary negotiations

A $5,000 raise when you expected nothing feels like a big gain. The same $5,000 when you expected $10,000 feels like a $5,000 loss. Skilled negotiators manage the reference point with ambitious opening anchors, because perception is relative.

7. Medical decisions

As the McNeil et al. (1982) study showed, “survival” and “mortality” framings of identical surgery and radiation-therapy data lead to different choices. The editing phase codes one frame as gains and the other as losses.

Prospect theory in your trading account

From my CFD trading on Plus500

I’ve caught myself doing exactly what the reflection effect predicts with gold positions. A $20 unrealised profit triggers the urge to close immediately (“lock it in!”). A $40 unrealised loss produces the opposite (“it’ll come back, the thesis is intact”). The result was cutting winners at 1R and letting losers run past 2R. Forcing mechanical exits with pre-set stop-loss and take-profit levels was worth more than any improvement in technical analysis.

Structural countermeasures beat willpower:

  • Pre-commit to exit rules. Set the stop-loss and take-profit before entering, while you are still at the reference point, then execute mechanically. Compare market, limit and stop orders.
  • Size positions so no single loss is painful. See position sizing and risk per trade.
  • Extend your evaluation horizon. Checking less often reduces the number of losses you experience (myopic loss aversion).
  • Think in expected value. A system with a 40% win rate and 3:1 reward-to-risk has +0.6R expected value per trade (0.4 × 3 − 0.6 × 1). Prospect theory predicts you’ll still feel bad about the 60% of losing trades and may abandon a profitable system. A trade journal that tracks running expectancy helps override that signal.

Investor sophistication helps but does not eliminate the bias: Dhar and Zhu (Management Science, 2006) found a weaker disposition effect among wealthier, more experienced and professional-occupation investors. See also behavioral finance: 7 biases that cost you money.

Prospect theory in Python: value and probability weighting

The snippet below implements the CPT value and weighting functions with Tversky and Kahneman’s 1992 median parameters. It reproduces three predictions: a fair coin flip is rejected, a long-shot lottery is preferred to its expected value, and insurance is preferred to the risk it covers.

Python
# Cumulative prospect theory - Tversky & Kahneman (1992) median parameters
ALPHA, BETA, LAMBDA = 0.88, 0.88, 2.25   # value function
GAMMA, DELTA = 0.61, 0.69                # probability weighting (gains, losses)

def value(x: float) -> float:
    """S-shaped value function, relative to the reference point (x = 0)."""
    return x ** ALPHA if x >= 0 else -LAMBDA * (-x) ** BETA

def weight(p: float, c: float) -> float:
    """Inverse-S probability weighting: overweights small p, underweights large p."""
    return p ** c / (p ** c + (1 - p) ** c) ** (1 / c)

def two_outcome_prospect(x: float, p: float) -> float:
    """Value of 'x with probability p, otherwise 0'."""
    c = GAMMA if x >= 0 else DELTA
    return weight(p, c) * value(x)

print(f"w(1%)  = {weight(0.01, GAMMA):.3f}")   # 0.055 - a 1% chance feels like ~5.5%
print(f"w(50%) = {weight(0.50, GAMMA):.3f}")   # 0.421
print(f"w(99%) = {weight(0.99, GAMMA):.3f}")   # 0.912

# Fair coin flip on $100: CPT weights gains and losses separately
flip = weight(0.5, GAMMA) * value(100) + weight(0.5, DELTA) * value(-100)
print(f"50/50 win or lose $100: {flip:.1f}")   # -34.6 -> rejected

# Lottery: 1% chance of $10,000 vs a sure $100 (same expected value)
print(f"lottery {two_outcome_prospect(10_000, 0.01):.1f} vs sure {value(100):.1f}")
# lottery 183.0 vs sure 57.5 -> risk-seeking for low-probability gains

# Insurance: 1% chance of losing $10,000 vs paying a sure $100 premium
print(f"risk {two_outcome_prospect(-10_000, 0.01):.1f} vs premium {value(-100):.1f}")
# risk -295.6 vs premium -129.5 -> risk-averse for low-probability losses

Does prospect theory still hold up? Evidence and criticisms

The core patterns replicate. Ruggeri et al. (Nature Human Behaviour, 2020) re-ran the original 1979 problems with 4,098 participants in 19 countries and 13 languages, changing only the currency. They replicated 94% of items and 12 of 13 theoretical contrasts, with effects somewhat smaller than in 1979 and real differences between countries. The authors concluded that the empirical foundations of prospect theory “replicate beyond any reasonable thresholds”.

The size of loss aversion is contested. Estimates of λ range from about 1 to 2.25 depending on the design. Brown et al. (2024) find a mean near 1.96 across 607 estimates. Walasek, Mullett and Stewart (2024) find about 1.31 in risky-choice data. The full debate is covered on our loss aversion page.

The editing phase is under-specified. The theory does not say exactly how people choose their reference point or frame, and that choice often decides the outcome. Parameters also vary widely between individuals, so the 1992 medians describe a “typical” chooser, not you.

Knowing the theory doesn’t cure the bias. The patterns come from fast, intuitive judgement and persist even when people understand them. Structural fixes — defaults, pre-commitment, fewer evaluations — work better than willpower. For a broader map of these patterns, see 12 cognitive biases that distort your decisions.

Prospect theory and AI in 2026

Two developments matter for anyone building or regulating AI products in finance. First, researchers now test large language models with prospect-theory tasks. Ross, Kim and Lo (COLM 2024) found that current LLMs are neither fully human-like nor fully rational in risk and loss-aversion tasks, and that most behave inconsistently across settings. A model advising users can carry its own framing sensitivity, so test gain-framed and loss-framed versions of the same question.

Second, regulation. Under the EU AI Act (Regulation (EU) 2024/1689), Article 5 has prohibited since 2 February 2025 AI systems that use purposefully manipulative or deceptive techniques, or that exploit vulnerabilities linked to age, disability or social or economic situation, to materially distort someone’s behaviour in a way that causes or is likely to cause significant harm. Deliberately exploiting loss framing or probability weighting in an AI-driven financial product could come under scrutiny on those grounds. Ordinary, transparent nudges are not banned. On the trading side, see how these tools are used in AI in trading. Polish readers: teoria perspektywy — 5 efektów.

FAQ

What is prospect theory in simple terms?

Prospect theory says people judge outcomes as gains or losses relative to a reference point rather than in absolute terms. They feel losses about twice as strongly as equal gains, take risks to avoid sure losses, and overweight small probabilities. It explains why the same person buys insurance and lottery tickets, and why investors sell winners too early.

What is Kahneman and Tversky’s prospect theory?

It is the model Daniel Kahneman and Amos Tversky published in Econometrica in 1979, “Prospect Theory: An Analysis of Decision under Risk”, as a descriptive alternative to expected utility theory. They refined it as cumulative prospect theory in 1992. Kahneman received the 2002 Nobel Memorial Prize in Economic Sciences and died in March 2024. Tversky died in 1996.

What is cumulative prospect theory?

Cumulative prospect theory (Tversky and Kahneman, 1992) is the updated version of the model. It applies probability weights to cumulative, rank-ordered probabilities instead of to each outcome separately, so it never violates stochastic dominance. It also handles any number of outcomes and uncertainty, and uses separate weighting for gains and losses. Its median parameters are α = β = 0.88, λ = 2.25, γ = 0.61 and δ = 0.69.

What is the reflection effect in prospect theory?

The reflection effect is the reversal of risk attitudes between gains and losses. In 1979, 80% of people chose a sure $3,000 over an 80% chance of $4,000, but 92% chose an 80% chance of losing $4,000 over a sure loss of $3,000. People are risk-averse for gains and risk-seeking for losses.

What is probability weighting?

Probability weighting means people treat probabilities non-linearly. They overweight small probabilities and underweight moderate and large ones. With Tversky and Kahneman’s 1992 parameters, a 1% chance feels like about 5.5% and a 99% chance like about 91%. It explains lottery buying, insurance buying and the certainty effect.

How is prospect theory related to loss aversion?

Loss aversion is one of the four building blocks of prospect theory, alongside reference dependence, diminishing sensitivity and probability weighting. It is captured by the coefficient λ, estimated at 2.25 in 1992 and around 1.3–2 in 2024 meta-analyses. Prospect theory is the full model; loss aversion is the part that makes losses loom larger than gains.

What is the difference between prospect theory and expected utility theory?

Expected utility theory is normative: it says rational people evaluate final wealth using objective probabilities. Prospect theory is descriptive: people evaluate gains and losses from a reference point, weight losses more, distort probabilities and respond to framing. Prospect theory explains behaviors such as the disposition effect and the four-fold pattern that expected utility struggles with.

References

  1. Kahneman, D., & Tversky, A. (1979). Prospect Theory: An Analysis of Decision under Risk. Econometrica, 47(2), 263–291. https://doi.org/10.2307/1914185
  2. Tversky, A., & Kahneman, D. (1992). Advances in Prospect Theory: Cumulative Representation of Uncertainty. Journal of Risk and Uncertainty, 5(4), 297–323. https://doi.org/10.1007/BF00122574
  3. Tversky, A., & Kahneman, D. (1981). The Framing of Decisions and the Psychology of Choice. Science, 211(4481), 453–458. https://doi.org/10.1126/science.7455683
  4. Quiggin, J. (1982). A Theory of Anticipated Utility. Journal of Economic Behavior & Organization, 3(4), 323–343.
  5. Ruggeri, K., et al. (2020). Replicating patterns of prospect theory for decision under risk. Nature Human Behaviour, 4, 622–633. https://doi.org/10.1038/s41562-020-0886-x
  6. McNeil, B. J., Pauker, S. G., Sox, H. C., & Tversky, A. (1982). On the Elicitation of Preferences for Alternative Therapies. New England Journal of Medicine, 306(21), 1259–1262. https://doi.org/10.1056/NEJM198205273062103
  7. Brown, A. L., Imai, T., Vieider, F. M., & Camerer, C. F. (2024). Meta-analysis of Empirical Estimates of Loss Aversion. Journal of Economic Literature, 62(2), 485–516. https://doi.org/10.1257/jel.20221698
  8. Walasek, L., Mullett, T. L., & Stewart, N. (2024). A meta-analysis of loss aversion in risky contexts. Journal of Economic Psychology, 103, 102740. https://doi.org/10.1016/j.joep.2024.102740
  9. Benartzi, S., & Thaler, R. H. (1995). Myopic Loss Aversion and the Equity Premium Puzzle. The Quarterly Journal of Economics, 110(1), 73–92. https://doi.org/10.2307/2118511
  10. Shefrin, H., & Statman, M. (1985). The Disposition to Sell Winners Too Early and Ride Losers Too Long: Theory and Evidence. The Journal of Finance, 40(3), 777–790. https://doi.org/10.1111/j.1540-6261.1985.tb05002.x
  11. Samuelson, W., & Zeckhauser, R. (1988). Status Quo Bias in Decision Making. Journal of Risk and Uncertainty, 1(1), 7–59.
  12. Barberis, N., Jin, L. J., & Wang, B. (2021). Prospect Theory and Stock Market Anomalies. The Journal of Finance, 76(5), 2639–2687. https://doi.org/10.1111/jofi.13061
  13. Liu, Y.-Y., Nacher, J. C., Ochiai, T., Martino, M., & Altshuler, Y. (2014). Prospect Theory for Online Financial Trading. PLoS ONE, 9(10), e109458. https://doi.org/10.1371/journal.pone.0109458
  14. Dhar, R., & Zhu, N. (2006). Up Close and Personal: Investor Sophistication and the Disposition Effect. Management Science, 52(5), 726–740.
  15. Ross, J., Kim, Y., & Lo, A. W. (2024). LLM economicus? Mapping the Behavioral Biases of LLMs via Utility Theory. COLM 2024. arXiv:2408.02784
  16. Kahneman, D. (2011). Thinking, Fast and Slow. Farrar, Straus and Giroux.
  17. European Parliament and Council. (2024). Regulation (EU) 2024/1689 — Artificial Intelligence Act, Article 5. https://eur-lex.europa.eu/eli/reg/2024/1689/oj
Exit mobile version