≈$2M DEFINITIVE CONTRACT HR001120C0041

About $1.95M to Smart Information Flow Technologies — how AI behaves on meeting situations unseen in training (DARPA, phase 1)

Department of Defense 2019-11-18 〜 2023-05-18

A definitive contract awarded by the Defense Advanced Research Projects Agency, described as the science of artificial intelligence and learning for open-world novelty, base phase 1. It addresses behaviour on meeting situations not anticipated at the time of training. Obligations are about $1,953,923 and outlays about $292,627, a ratio of 15.0 percent.

Contract key facts

  • RecipientSMART INFORMATION FLOW TECHNOLOGIES LLC
  • Contract value$1,953,923 (≈$2M)
  • Awarding agencyDepartment of Defense
  • Awarding sub-agencyDefense Advanced Research Projects Agency
  • Award typeDEFINITIVE CONTRACT
  • Period of performance2019-11-18 〜 2023-05-18
  • Contract ID (PIID)HR001120C0041

Contract scope (original)

SCIENCE OF ARTIFICIAL INTELLIGENCE AND LEARNING FOR OPEN-WORLD NOVELTY (SAIL-ON) - BASE (PHASE 1)

Key points

  • Open-world novelty means meeting a situation not anticipated at the time of training.
  • Separate from a performance figure, the subject is whether a system breaks outside what was anticipated or changes how it behaves.
  • The description states this is the base, phase 1, so the research is contracted in stages.
  • Obligations are about $1,953,923 and outlays about $292,627, a ratio of 15.0 percent.
  • Across the federal spending records this site holds as of 2026-09-01, 130 of the 155 Department of Defense awards show zero outlays.

1The phrase open world

Open-world novelty means meeting a situation that was not anticipated at the time of training. Rather than performing within a fixed range, the subject is what happens on stepping outside it.

A model is strong on inputs close to what it learned and weak on inputs far from them. In actual use the outside of what was anticipated always turns up. Whether a system breaks there, or notices and changes how it behaves, is a question separate from a performance figure. This research takes that as its subject.

2The stage is written into the name

The description states plainly that this is the base, phase 1. A research plan proceeds in stages, with a contract cut per stage.

Dividing into stages allows a decision on whether to continue after seeing interim results. In the record, though, one piece of research appears as several contracts. Judging the scale of the research from the amount alone understates it.

3An outlay ratio of 15.0 percent

Obligations are about $1,953,923 and outlays about $292,627, a ratio of 15.0 percent. The period ran from 2019-11-18 to 2023-05-18, 3.5 years, and has ended.

Outlays against obligations292,627 / 1,953,923

Across the 222 federal spending records this site holds as of 2026-09-01, 130 of the 155 awarded by the Department of Defense show zero outlays. Records like this one carrying a non-zero value are a minority. Whether a value appears in the outlay field turns more on the reporting arrangement than on the state of the contract.

Zero-outlay records (Department of Defense)130 / 155

Why it matters

Performance is usually measured inside an anticipated range. What causes trouble in use lies outside it, and behaviour there has to be checked separately. Contracting in stages allows judgment along the way while making the scale of research harder to see from outside.

FAQ

What is an open world?
A situation where things not anticipated at the time of training appear, setting aside the premise of operating only within a fixed range.
Why divide into stages?
It allows a decision on whether to continue after seeing interim results. In the record one piece of research then appears as several contracts.
Is this the size of the whole research?
No. It covers phase 1, and later stages may exist as separate contracts, as the description notation indicates.

Sources (primary)

This article is an independent organization based on the U.S. official spending data below. Verify the exact, latest details with the official source.

#Federal spending#Contracts#DARPA#Machine learning#Robustness
Disclaimer: This site independently summarizes and classifies information based on official data sources. Always verify the latest and accurate information with the official sources. Content on finance, health, legal, and security is information, not advice. This site is not an official website of the U.S. government.