April 3, 2020 | International, Aerospace
Vice-président recherche, technologie, innovation du groupe Thales, David Sadek intervenait jeudi 24 octobre à Toulouse au Forum Innovation IA, un événement organisé par Sciences et Avenir pour le groupe Challenges.
Explicabilité, validité, intégrité... " Pour une IA de confiance " : c'était l'intitulé de l'intervention de David Sadek, le vice-président recherche, technologie, innovation du groupe Thales. "Nous aimerions tous pouvoir toujours faire confiance à tous nos outils technologiques. Mais en réalité tout dépend du contexte. Dans le cas des outils reposant sur l'IA, si Netflix me recommande un film que je trouve détestable, ce n'est pas bien grave, on s'en remettra. Mais dans le cas des systèmes critiques tels que l'on en conçoit régulièrement chez Thales pour des applications défense, espace et aéronautique, la question de la confiance est primordiale".
David Sadek énonce trois impératifs, qui sont autant d'axes stratégiques de la feuille de route IA chez Thales, pour aller vers cette " IA de confiance ". Pour qu'une application de l'IA soi digne de confiance, il faut d'abord qu'elle soit capable d'expliquer. Si elle se contente de proposer ou décider sans être capable de dire pourquoi et comment elle est arrivée à cette proposition ou décision, la défiance s'installera. Elle doit donc être capable de répondre à la question : " Pourquoi ? ".
"Si par exemple un copilote IA recommande au pilote humain de virer à 45°, et que ce dernier demande " pourquoi ", le copilote doit pouvoir répondre par exemple : " parce qu'il y a un problème météo (ou une menace ...) droit devant ". Il ne s'agit pas de simplement tracer la décision, de pouvoir indiquer à ses concepteurs quels neurones, quelles couches de neurones, ont fait pencher la balance dans un sens ou un autre, mais bien d'expliquer dans une langue compréhensible immédiatement par le pilote". Ce qui indique qu'à l'aspect intelligence artificielle s'ajoute celui de " l'interaction homme-machine ".
Cela implique, en restant sur cet exemple, une capacité de compréhension et de génération du langage naturel. Mieux, le pilote appréciera de pouvoir dialoguer dans un langage mixte, reposant sur l'oral et le geste (pour désigner un point sur une carte, etc.). Deuxième impératif selon David Sadek : il faut pouvoir démontrer la validité d'une application de l'IA. C'est-à-dire sa conformité aux spécifications. "Le système développé doit faire tout ce l'on attend de lui et rien que ce que l'on attend de lui. Ce qui suppose que l'on sache spécifier très proprement". On connait l'exemple d'un jeu de bataille navale, un " serious game " destiné à former des officiers de la marine, qui gagnait à tous les coups, parce qu'il sacrifiait systématiquement chaque vaisseau touché, pour ne pas ralentir la flotte.
Aucune règle ne prévoyait, n'interdisait ce cas de figure, passablement inacceptable. Troisième impératif pour une " IA de confiance ", selon David Sadek, la responsabilité des systèmes reposant sur l'IA. C'est-à-dire leur conformité aux cadres légaux, réglementaires et moraux. On connait le tendon d'Achille des réseaux de neurones : ce sont des boîtes noires, on voit ce qui entre et ce qui sort, mais on ne sait pas (trop) ce qu'il se passe à l'intérieur. C'est pourquoi David Sadek insiste sur le fait qu'il faut s'intéresser aussi à " l'autre IA ", l'IA symbolique, celle qui repose sur des règles et des raisonnements et qui s'oppose à l'IA connexionniste des réseaux de neurones. Cette IA symbolique, qui fait moins les gros titres, qui n'a pas connu les progrès fulgurants de l'IA des réseaux de neurones, est plus à même de répondre à ces trois exigences : expliquer, valider, responsabilité. C'est pourquoi la réponse à cette triple exigence passe sans doute, entre autres, par " l'IA hybride ", une IA tirant parti de ces deux branches de l'intelligence artificielle.
Par Pierre Vandeginste
https://www.sciencesetavenir.fr/high-tech/intelligence-artificielle/ia-pvdg_138526
October 21, 2023 | International, Aerospace
Spanish rocket company PLD Space plans a first orbital launch from French Guiana in the first quarter of 2026, after it carried out the first fully private European rocket launch earlier this month, it said on Friday.
September 17, 2018 | International, Aerospace
By: Valerie Insinna WASHINGTON — A fledgling five-person software start-up has just wrapped up their pitch to what could be a key investor. The pressure has been intense, but the customer decides to bite, sliding a credit card through a Square reader to award money to the company's Paypal account. This isn't an episode of Shark Tank, or a successful bid to get investment dollars from a Silicon Valley angel investor. The customer in this scenario is the Air Force — or at least, its acquisition executive would like it to be. During the Air Force Association's annual conference, Will Roper, assistant secretary of the Air Force for acquisition, technology and logistics, will announce a series of “Startup Days” targeted toward involving startup companies in Air Force acquisition by rapidly awarding contracts in less than 24 hours. “From the company's view, it will be a single day. From our view, we could probably turn it as quickly as a few days to a week — we put a call out, someone submits their idea, we analyze their idea, we check their company profile, we then invite a subset that have met the criteria,” he told Defense News in an exclusive Sept. 7 interview. “But then, when they come in to pitch their idea, they have a reasonable expectation of leaving with funding. We're ready to go or not go on a single day,” he continued. “And the contract length? One page.” When Defense News spoke with Roper, he had just gotten the green light to move ahead with this new way of awarding contracts. The effort remains in its early stages, and dates for Startup Day have not been chosen, although Roper believes the service could hold a series of Startup Days as soon as the end of this year or early 2019. The exact format is still being worked out as well: Program managers will be able to suggest precise technical problems that they'd like to see solutions for, but he'd also like to give companies the latitude to pitch their products for requirements the Air Force doesn't even know it has. Companies will submit proposals, which will be evaluated by Air Force program and contacting officials who will also analyze the company's profile — its number of employees, business type, product maturity and potential impact. But the goal is to have the actual events structured like a meeting with an angel investor, not the typical PowerPoint-laden gatherings of military officials and defense primes. “We've got to make this look more like Kickstarter than a defense industry day,” Roper said. “We may even put them on contract swiping a Square reader. We have government purchase cards that we're able to use for small purchases — up to $150k per transaction. That may be the mechanism we use because most companies that are startups, I'm going to guess, have a Paypal account.” An industry day geared specifically for start-up companies is just the latest way the Air Force is trying to harness a commercial technology boom where innovation has often been led by startups. Last year, the service announced the creation of a new organization called AFWERX that it began to help engage elements of the private sector that don't usually work with the government. However, even with AFWERX in operation, it takes the Air Force six to eight weeks to award a contract at its very fastest. And that's still too slow of a pace to enable it to work effectively with startup companies, Roper said. “There's this artificial ceiling that small companies can't reach to work with the government simply because they're too small to wait for a paycheck. If they're not on contract with us now, they've got to work with an investor fast enough to fund cash flow rates that startups need to grow,” he said. Roper doesn't expect all investments to bear fruit, but efforts like Startup Day have other advantages, he said. It gets Air Force contracting officers more comfortable with executing rapid contracts, potentially gives program managers a more effective way of spending their small business dollars and allows the service to have a voice in the kinds of technologies that cutting-edge companies develop. “I hope that will mean that every year when we do this — if it's successful, we'll do it every year—companies will have us on their radar screen and think, ‘The Air Force is a great way for us to get from being a company of five to a company of 50...and then we'll go off and become billionaires working with Amazon and Google,'” he said. “But this way they'll know us and their products and projects will have been influenced by us, hopefully for the betterment of the Air Force.” https://www.defensenews.com/digital-show-dailies/air-force-association/2018/09/17/air-force-aims-to-move-startups-from-pitch-to-contract-award-in-24-hours
December 4, 2019 | International, Aerospace, C4ISR
By: Valerie Insinna ORLANDO, Fla. — As the U.S. military prepares for the release of its fiscal 2021 budget request, Air Force leaders have made clear that a massive financial hike is needed for multidomain command-and-control efforts to connect weapon systems across the joint force. As part of that initiative, it will also be critical for the military to link together simulators so service members can replicate combat on a massive scale, Air Force Vice Chief of Staff Stephen Wilson said during a speech Tuesday at the Interservice/Industry, Training, Simulation and Education Conference. “I know you spent a lot of time yesterday about how to connect and integrate simulators from across our various weapons systems. I can't overstate how important that is,” he said. “Individual weapon system simulators can help our men and women become proficient tacticians, but it's their ability to integrate and connect that will differentiate us against a peer threat. And no one is going to win alone." Wilson said the Air Force plans to spend “a lot of money in this next five-year defense plan” on multidomain command and control, but he did not detail how much funding the service might ask for, or whether that sum will include investments in technology to network together simulators. The service is still in the earliest stages of identifying how to connect its aircraft and space assets with the joint force to fight advanced, near-peer threats like Russia and China. The same goes for its aircraft simulators, which are largely detached from each other. An Air Force program called Simulator Common Architecture Requirements and Standards, or SCARS, will start to transform the simulation enterprise and make it more interconnected, said Col. Phillip Carpenter, the Air Force's senior materiel leader for the simulators program office. “It's an effort to make the entire portfolio more modular, more open-system,” he told Defense News in an exclusive interview. “I'm not going to say [SCARS] is the solution to all of our problems,” but it will “help lay that groundwork so that we become much more interoperable across the board.” Under SCARS, the Air Force wants to create a common, open architecture for its simulators that will impose stricter cybersecurity standards and make it easier for the service to update simulators with new capabilities or threat information. The goal, Carpenter said, is to have a fleet of simulators that can remotely receive software updates, much like a smartphone. The Air Force released a SCARS request for proposals in December 2018 for a 10-year contract worth up to $900 million, according to Bloomberg. The service intends to award a contract for SCARS in 2020 and has received a lot of interest from industry, Carpenter said. Carpenter was clear that SCARS — at least how it is currently conceived — will not enable the Air Force to carry out the type of scenarios Wilson spoke about: large-scale, simulated air operations involving simulators of many different airframes. To achieve that, “we need to work on some of the security pieces so that we can allow multilevel security or some of these other aspects that would allow totality of these systems to participate and fight like they would actually fight in a real world,” Carpenter explained. But Carpenter believes there is potential for other parts of the Air Force — like its space, cyber and intelligence community — or even the other services to adopt SCARS or an architecture that is compatible with it. “I think there is great potential for SCARS to be more than something for just aircraft simulators,” he said. “If somebody is off building a system outside of our portfolio, if it's built to a common standard, I think that would effectively make all the systems, whether they're in our portfolio or not, more interoperable.” https://www.defensenews.com/digital-show-dailies/itsec/2019/12/03/the-us-air-force-wants-to-network-all-its-weapons-together-will-simulators-be-included