Cross-posted from
https://forum.arctic-sea-ice.net/index.php/topic,776.msg223266.html#msg223266The US Military’s AI Can’t Find Targets On Its Own — Yet, Top USAF General Sayshttps://www.defenseone.com/technology/2019/08/ai-cant-find-targets-their-own-yet-top-usaf-general-says/159313/The Air Combat Command leader says the tools are still learning.Nearly two years since the Pentagon started bringing artificial intelligence to the battlefield, the algorithms still need human help, a top U.S. Air Force general said Tuesday.
But Gen. Mike Holmes said
the technology is getting better at identifying people, cars, and other objects in drone video.“[W]e’re still in the process of teaching the algorithms to be able to predict what’s there from the data and be as reliable as we would like it to be or as reliable as our teams of people [who] are doing that,” the Air Combat Command leader said Tuesday at a Defense Writers Group breakfast.
... “Those tools are there. We’re starting to use them and experiment with them,” he said. “I don’t think, in general, they’re at the point yet where we’re confident in them operating without having a person following through on it, but I absolutely think that’s where we’re going.”In June 2018,
Holmes called artificial intelligence “a big part of our future and you’ll continue to see that expanded.”
Asked by a reporter if Project Maven is the first step toward Skynet, a fictional artificial network in the Terminator movies, Holmes said: “I certainly hope not.”
-----------------------------
America Needs a “AI Dead Hand”https://warontherocks.com/2019/08/america-needs-a-dead-hand/Rational? Rational? ....
To maintain the deterrent value of America’s strategic nuclear forces, the United States may need to develop something that might seem unfathomable — an automated strategic response system based on artificial intelligence.Admittedly, such a suggestion will generate comparisons to Dr. Strangelove’s doomsday machine, War Games’ War Operation Plan Response, and the Terminator’s Skynet, but the prophetic imagery of these science fiction films is quickly becoming reality. A rational look at the NC
3 modernization problem finds that it is compounded by technical threats that are likely to impact strategic forces.
Time compression has placed America’s senior leadership in a situation where the existing NC
3 system may not act rapidly enough.
Thus, it may be necessary to develop a system based on artificial intelligence, with predetermined response decisions, that detects, decides, and directs strategic forces with such speed that the attack-time compression challenge does not place the United States in an impossible position.New technologies are shrinking America’s senior-leader decision time to such a narrow window that it may soon be impossible to effectively
detect, decide, and direct nuclear force in time. In the wake of a nuclear attack, confusion and paralysis by information and misinformation could occur when the NC3 system is in a degraded state.
It is conceivable that attack-time compression will reorder the process: the president will decide ahead of time what response will take place for a given action and it will then be left to artificial intelligence to detect an attack, decide which response is appropriate (based on previously approved options), and direct an American response. Such a system would differ significantly from the Russian Perimeter system since it would be far more than an automated “dead man” switch — the system itself would determine the response based on its own assessment of the inbound threat.... There is a[n] option. The United States could develop an NC3 system based on artificial intelligence. Such an approach could overcome the attack-time compression challenge.DARPA’s Knowledge-directed Artificial Intelligence Reasoning Over Schemas program is an example of how an American NC3 system based on artificial intelligence might function. Fusing the contextual and temporal events of a nuclear attack into an analytic-based artificial intelligence capability may ensure rapid comprehension and in turn generate associated and prompt actionable responses. The biggest challenge for such a system is its ability to learn and adapt.
Artificial intelligence is already being used for target identification, controlling autonomous platforms, pattern recognition, and a number of other wartime tasks.
It is capable of processing vast amounts of information very quickly and assessing the pros and cons of alternative actions in a thoroughly unemotional manner. --------------------------
Meet the Classified Artificial Brain Developed by US Intelligence Programs: IT’S SENTIENThttps://www.theverge.com/2019/7/31/20746926/sentient-national-reconnaissance-office-spy-satellites-artificial-intelligence-aiUntil now, Sentient has been treated as a government secret, except for vague allusions in a few speeches and presentations. But recently released documents — many formerly classified secret or top secret — reveal new details about the program’s goals, progress, and reach.
... The agency has been developing this artificial brain for years, but details available to the public remain scarce. “It ingests high volumes of data and processes it,” says Furgerson. “Sentient catalogs normal patterns, detects anomalies, and helps forecast and model adversaries’ potential courses of action.” The NRO did not provide examples of patterns or anomalies, but one could imagine that things like “not moving a missile” versus “moving a missile” might be on the list. Those forecasts in hand, Sentient could turn satellites’ sensors to the right place at the right time to catch ill will (or whatever else it wants to see) in action. “Sentient is a thinking system,” says Furgerson. (
... think Echelon cubed)
We don’t know, exactly ... which sorts of data sources Sentient may siphon in, but it’s clear that the program is interested in all kinds of information. Retired CIA analyst Allen Thomson goes further. “As I understand it, the intended — and aspirational — answer is ‘everything,’” he says. In addition to images, that could include financial data, weather information, shipping stats, information from Google searches, records of pharmaceutical purchases, social media and more, he says.
... The NRO notes that Sentient doesn’t keep people totally out of the process, providing some kind of check on its state of being. “Having humans in the loop overseeing the intelligence data and information is a key way of monitoring the algorithm’s performance,” says Furgerson. “Sentient is human-aided machine-to-machine learning.”
------------------------------
Office of the Director of National intelligence: Quadrennial Intelligence Community Review
See pg 6: Sentient AIhttps://theintercept.com/document/2014/09/05/quadrennial-intelligence-review-final-report-2009/… (C//REL) ... The Sentient Enterprise will track and manage thousands of exabytes of data every day (1 exabyte is the equivalent of 100,000 times the Library of Congress, which holds 19 million books), enabling iterative assessments in real time, not days or weeks. The data it manages will be universally discoverable, accessible, and usable by humans and machines equally. Indeed, the human-machine interface will allow the individual to interact directly with a unified information architecture. The enterprise will be able to continuously and autonomously process, evaluate, and act on new data without regard to structure or format. The enterprise will log expert users’ interactions with the data, while gleaning new insights from more generalist users. By so doing, the entire enterprise will create, share, and advance corporate knowledge in a rich and seamless interplay where machines and humans learn together.
(U) Sample Key Capabilities
• (C//REL) Automation. The IC would emplace sensors and monitor applications that run autonomous collection of the most relevant data, trigger pattern recognition sequences, and process raw feeds. This would require supercomputer-like capabilities at every “computational point-of-presence,” from computer terminal to digital handheld device. Automation (e.g., in language translation, gisting, relational analysis, and trend assessment) would allow verification and validation of the accuracy of information.
• (U) Artificial Intelligence (AI). Application of advanced AI techniques would make it possible to continuously improve understanding of complex threat environments, discern the relative importance of data, and adapt quickly to changes indicated by sensor data and automated analysis (thus providing indication and warning). This would allow the experts to focus on translating critical information to decision-makers in an effective and time-efficient manner.
• (S//REL) Self-Learning. The institutional knowledge of the Sentient Enterprise would increase the user’s ability to recall events and significant facts to build relational awareness. Simultaneously, the human-machine interface would enable the user to continuously refine the “algorithms” that translate human judgments into machine language so that the system actively learns.https://www.nro.gov/Portals/65/documents/foia/declass/ForAll/051719/F-2018-00108_C05113686.pdfhttps://www.nro.gov/Portals/65/documents/foia/declass/ForAll/051719/F-2018-00108_C05112980.pdfhttps://www.nro.gov/Portals/65/documents/foia/declass/ForAll/051719/F-2018-00108_C05113682.pdf