# Firm ground By Ali Khalilli. Deterministic Labs, August 2026. Canonical: https://deterministiclabs.ai/writing/firm-ground Why our machines sound right when they are wrong, and what to do about it. On the kinds of problems, the layers of certainty, and where to build. A clock that runs exactly one hour fast is never right. Not once a day, not once a year; it is always one hour ahead of the truth. A stopped clock, at least, is right twice a day. Both clocks are perfectly consistent, each does the same thing every time we look. But: one is sometimes right and one never is. So consistency and correctness are different things. A machine can be perfectly consistent and perfectly wrong at the same time. If we look at which clock is the dangerous one? The stopped clock fools nobody: anyone can see it is dead. The running clock ticks, spins, and looks alive. And it is wrong every single time. Stopped right twice a day One hour fast never right Figure: Each shows the same time whenever we look. Both are perfectly consistent; only one is sometimes correct, and it is the one that looks dead. AI today is becoming that running clock. The trick is old, the scale is new. And there is a reason it works on us. Psychologists describe two modes of human judgment: a fast one that runs on impressions, and a slow one that checks[1]questionHow many months are in a year? Figure: Twelve, read at five heights, drawn as an iceberg: solid is locked, open water is free. Ask again and the free layers vary while the locked layers repeat; every run leaves a receipt. Tap a layer to lock or free it. The same answer can be locked on some layers and free on others, at once. That gives the whole craft in one sentence: **decide which layers must not change, lock them, and let the others stay free.** For the number twelve, lock the meaning and free the wording. An AI forbidden to say “a dozen” is not more precise. It is worse at language. The reverse works too: a report can have lively writing and exact numbers at the same time. The style is free. The math is locked. Different rules on different layers, inside one answer. Most arguments about whether AI is “reliable enough” are really two people talking about two different layers without noticing. ## Where the randomness enters People say AI model is random. The truth is more precise, and more useful. The model itself is fixed. Give it the same question twice and it produces the same internal result twice: a list of possible next words, each with a probability. Same list, same numbers, every time. The randomness enters in one extra step at the end. The system picks from that list partly by chance, like a weighted roll of dice. The chance is there on purpose: always choosing the top word makes the writing flat and repetitive[4]The questionevery output is a ________.ask×20↓The model: fixed0.31846 -0.00421 1.20894 -0.55210 0.66702 -0.28451 0.09327 -1.14082 0.52290 -0.07361-0.72913 0.66048 -0.09134 0.42718 -1.05263 0.18095 -0.44816 0.91037 -0.23568 0.701540.10529 -0.84306 0.37715 1.08841 -0.51172 0.29408 -0.90253 0.06147 1.17520 -0.38096-0.15540 0.28093 -0.97625 0.63410 0.04821 -0.66539 1.02174 -0.30988 0.47206 -0.85417the modelasked ×0, changed ×0↓The list: mattergroup orderpinned↓Choice: the dicea weighted roll, on purposeroll the dicetop word↓The answerclear ×________not asked yetThe model computes the same list on every prompt. The spread we see is the dice, rolled on purpose. But a second randomness hides in the last decimals: other people’s requests, running at the same time, change the order the sums are added in. Take the top word and pin the order, and the same question returns the same answer, every time.We tell this small story because it is the whole domain in one small example. Exactness is not something computers give us for free. It is something we build, after we understand precisely what was moving and why. “Computers are just random sometimes” is a give up & excuse. “This sum, in this order, pinned down” is firm ground. One clarification, before the bigger idea. Deterministic does not mean predictable. The weather follows fixed physical laws, and still nobody can forecast it three weeks ahead, because tiny differences today grow into huge differences later[7]stonefogdunepress the sky to askbuild in stonename the condition firstan honest opinion, labeledpress the sky to askis 2,147,483,647 prime? at what temperature does water boil? the fog asks back: at what altitude? is this person tall? Figure: One sky, three grounds. On stone every drop lands on the same exact line. In fog the stone dissolves until you name the hidden condition; drag the altitude and the answer turns exact. On the dune each drop settles differently, an honest opinion against the standard you chose. The guarantee must match the ground. You would not trust a calculator that got creative. You would not want a poet who wrote the same poem every time. The guarantee must match the ground. And this is the failure we are here to fight: treating soft ground as solid. Pouring concrete over mud, stamping it “verified,” and selling it with the confidence of the fast clock. The opposite failure is real too: treating solid ground as soft, and answering a question of plain fact with a vague opinion. Both failures come from the same blindness. Not knowing, or not caring, what kind of ground you stand on. ## The line moves One more fact, and it is the most important one here. The line between “checkable” and “matter of opinion” is not fixed. People move it, deliberately, by building tools. Whether a mathematical proof was correct used to be an expert’s opinion. Reviewers read it and judged, and reviewers disagreed. The four-color theorem was contested for years because its original 1976 proof used a computer that no human could fully audit. Then proofs were given a form that programs can verify, and a machine now checks the entire theorem, every line, in minutes[9] This is the second meaning of our name. Deterministic is not only what we look for. It is what we do: we make things checkable that were not. There are many questions in AI today that receive a confident opinion when they could receive a check. Our work is moving them, one by one, from “trust me” to “check it.” ## What we know, and what we bet An essay that asks machines to label their claims should label its own. What we know: consistency and correctness are different things. “The same answer” has layers, and each layer can be locked separately. The randomness in AI comes almost entirely from one sampling step, plus a small hardware quirk, and both can be controlled. And questions come in three kinds, each bearing a different kind of certainty. What we bet: the industry is perfecting the sound of intelligence, and the sound is nearly perfect. Our first bet is about skill. The scarce ability of the coming years is not fluency but discrimination: knowing what kind of question is on the table, and answering with the kind of certainty that question can carry. Our second bet is about what follows from that skill. Trust will reorganize this market. Systems that can be checked will be given decisions. Systems that must be believed will be given suggestions. The distance between those two roles is where the value of machine intelligence will settle. We may be wrong. That is what makes them bets. ## Where we stand Today’s AI answers a calculation, an estimate, and a matter of taste in one identical, confident voice, and gives you no way to tell which is which. That is the fast clock at full scale: a machine that is wrong in the same tone it uses when it is right. And the tone is not an accident. These systems are tuned on human approval, and human approval follows the fast judgment: we reward what sounds right[11] Its answers are made to be believed. Ours are made to be checked. Everything we release will arrive with the means to check it. Run it yourself. Measure it yourself. Compare it yourself. We begin at the bottom. From there we build upward, layer by layer, and each layer will stand on a checked one. For us, determinism means three things. Knowing where certainty is real. Being honest about where it is not. And moving the line between them. It is the difference between the clock that is confidently never right, and the one that knows its limits and keeps time. Everything we make begins here. ## References - Kahneman, D. *Thinking, Fast and Slow.* Farrar, Straus and Giroux, 2011. - Alter, A. L., and Oppenheimer, D. M. “Uniting the Tribes of Fluency to Form a Metacognitive Nation.” *Personality and Social Psychology Review,* 13(3), 219–235, 2009. doi.org/10.1177/1088868309341564 - “Deterministic.” *Cambridge Dictionary.* dictionary.cambridge.org/dictionary/english/deterministic - Holtzman, A., Buys, J., Du, L., Forbes, M., and Choi, Y. “The Curious Case of Neural Text Degeneration.” *ICLR,* 2020. arxiv.org/abs/1904.09751 - Goldberg, D. “What Every Computer Scientist Should Know About Floating-Point Arithmetic.” *ACM Computing Surveys,* 23(1), 5–48, 1991. doi.org/10.1145/103162.103163 - He, Horace, and Thinking Machines Lab. “Defeating Nondeterminism in LLM Inference.” *Thinking Machines Lab: Connectionism,* September 2025. thinkingmachines.ai/blog/defeating-nondeterminism-in-llm-inference - Lorenz, E. N. “Deterministic Nonperiodic Flow.” *Journal of the Atmospheric Sciences,* 20(2), 130–141, 1963. journals.ametsoc.org/view/journals/atsc/20/2/1520-0469_1963_020_0130_dnf_2_0_co_2.xml - Dickson, L. E. *History of the Theory of Numbers, Volume I: Divisibility and Primality.* Carnegie Institution of Washington, 1919. Records Euler’s 1772 proof of the primality of 2³¹ − 1. archive.org/details/historyoftheoryo01dick - Gonthier, G. “Formal Proof: The Four-Color Theorem.” *Notices of the American Mathematical Society,* 55(11), 1382–1393, 2008. ams.org/notices/200811/tx081101382p.pdf - Leroy, X. “Formal Verification of a Realistic Compiler.” *Communications of the ACM,* 52(7), 107–115, 2009. xavierleroy.org/publi/compcert-CACM.pdf - Sharma, M., et al. “Towards Understanding Sycophancy in Language Models.” *ICLR,* 2024. arxiv.org/abs/2310.13548