The request for information was recently posted to the CDAO’s Tradewinds website.

The office has been tasked with developing a “data integration layer,” or DIL, to help the Pentagon achieve its vision for connecting the various sensors, shooters and information streams of the U.S. military services, allies and partners under a unified network. Officials refer to the concept as Combined Joint All-Domain Command and Control (CJADC2).

“Warfighters continue to require the ability to maintain and improve the decision advantage through using a data centric approach for successful battlefield outcomes. DoD warfighters must be able to dynamically integrate and operate as one truly joint force. To inform possible future acquisitions and improve our operational advantage, the Department seeks information related to technologies designed to improve data visibility, simplify data connections, and automate data access for legacy and new globally distributed warfighting systems supported by the DIL,” the RFI states.

The Chief Digital and AI Office is eyeing commercial technologies that can be folded into a zero-trust cybersecurity model, which all Defense Department agencies are expected to implement by 2027.

“The CDAO considers a data mesh, implemented with zero trust, with compliments of open source, non-proprietary & proprietary software components, as potential implementation of the CJADC2 DIL target architecture. The CDAO is currently experimenting with these technologies to inform the Data Integration Layer concept and requirements. The CDAO seeks industry’s input on the current state of commercial data-mesh capabilities to inform the CJADC2 Data Integration concept, governance, and technical capabilities,” per the RFI.

A key goal of the office is to provide a globally connected but decentralized mesh services architecture with a strong data backbone that can operate in environments where communications are denied, disconnected, intermittent or limited.

The envisioned data integration layer is expected to support the rapid deployment and integration of apps, “algorithmic services” and AI tools; secure role and access-based identity management and controls to facilitate data exchanges with international partners; and object and entity distribution services delivered across multiple domains and network types using data tagging, parsing, certificate, Blockchain or “other accreditable technologies,” according to the CDAO.

“Ancillary services such as data management tools for the operation of resilient data mesh services, data ingest [and] curation, and source to end point security will be considered essential,” the RFI states. “This request will serve to inform the government’s technical and operational understanding of such a capability and how it might be acquired, deployed, sustained, and modernized.”

The office is asking industry for information about how their data-mesh capabilities can work across multiple classification levels and security domains; enable “fault tolerant” software architectures; be synchronized between cloud regions, cloud geographies, cloud providers, and on-prem resources in both connected and denied, disconnected, intermittent or limited environments; enable the transmission of high-priority data ahead of lower-priority data; and meet other needs.

Industry responses are due Nov. 28.

Lawmakers are pressing the organization to make progress with the initiative. The House version of the fiscal 2024 defense policy bill would require the CDAO, no later than March 1, to provide a plan and associated timelines for deploying and demonstrating a joint data integration layer prototype in the U.S. Indo-Pacific Command area of operations, transitioning the technology to fielding by Indo-Pacom, and scaling that capability across the other combatant commands.

The Indo-Pacific is now the U.S. military’s top priority theater amid concerns about China’s military modernization and geopolitical competition in the region. The Defense Department wants to deploy cutting-edge technologies there to improve its force posture and operational prowess.

Congress has not yet passed the final version of the policy legislation.

The post Pentagon’s CDAO queries industry about commercial data-mesh capabilities appeared first on DefenseScoop.

The CDAO’s request for “discovery papers,” issued last week, comes as the Pentagon is pursuing what it calls “responsible AI,” including safeguards to mitigate the risk that algorithm-armed tools and weapons will go off the rails.

“The DoD is interested in supporting grassroots/crowdsourced red-teaming efforts to ensure that their AI-enabled systems — and the contexts in which they run — are safe, secure, reliable, and equitable. Bias — the systematic errors that an AI system generates due to incorrect assumptions of various types, is a threat to achieving this outcome. Therefore, as part of this priority, the current call seeks industry partners to help develop and run an AI bias bounty program to algorithmically audit models, facilitate experimentation with addressing identified risks, and ensure the systems are equitable given their particular deployment context,” according to a notice posted on the CDAO’s Tradewind website.

The Defense Department has been using so-called bug bounty programs, such as Hack the Pentagon, that offer financial awards to crowdsource efforts to discover cyber vulnerabilities. Now, it aims to apply a similar model to look for flaws in artificial intelligence, with an initial focus on large language models or generative AI-based systems.

“The DoD needs an approach to bias bounty programs that can be adapted to various technologies and contexts and scalable to DoD-wide. For the initial program, the DoD needs a vendor to set up, run, and coordinate the initial bounty. This includes adapting the target to the bounty task (as needed), socializing information about the bounty, providing technical support, gathering, vetting, and managing researchers, and evaluating and testing the submissions that identify sources of bias and/or experiment with solutions or submit improvements. Finally, the vendor should suggest approaches to successfully mitigate or address the identified biases,” per the notice.

The contract may be extended for the vendor to run bias bounty programs for “other types of models, use cases, and settings.” However, after work under the contract wraps up, the Pentagon expects the vendor to have equipped the DOD and its components with the tools needed to organize and run their own bias bounty programs in the future, according to the CDAO.

Discovery papers are due July 21. Offerors who advance to the second round after submitting their papers will enter a “digital proving ground” where they will have an opportunity to make their pitch to contracting officers looking to issue rapid pilot project awards using other transaction authority. The Pentagon may issue one award, multiple awards, or no awards, as a result of these engagements.

The post Pentagon looking for contractors to set up AI ‘bias bounty’ programs appeared first on DefenseScoop.

The open call, posted on the office’s Tradewind marketplace, comes as the U.S. military is making a push to develop and field “responsible AI” capabilities and autonomous platforms and weapons. Officials also recognize the need for warfighters to have confidence that the technology will work as advertised on the battlefield, or else they won’t be inclined to use it to its full potential.

The DOD “needs a comprehensive way to measure user trust in AI-enabled systems — including how trust may break down into various dimensions and its relationship to other concepts or constructs. For instance, dimensions of trust might include the extent to which an individual has confidence in the system’s accuracy vs. the extent to which an individual trusts the system because they hope that it will work; under vs. over-reliance on the system; etc. Related concepts or constructs include whether a user’s degree (and kind) of understanding of how the system functions contributes to their trust in the system; whether distrust is a separate construct from trust (vs. whether it is on the same scale); etc.,” the Tradewind announcement states.

“The assessment must enable a rich understanding (and prediction) of the descriptive psychological and behavioral states of users interacting with the system. The test must also provide sufficient data to support normative evaluations about the level of trust in the system, such as whether the trust in the system is justified given the evidence, or whether users over-trust the system in a way that impedes meaningful degrees of human judgment and control,” it notes.

The CDAO program aims to develop metrics and tests for assessing and evaluating user trust in AI-enabled systems; enable the continuous monitoring of operational users’ trust; and facilitate assessment of trust “among various dimensions and its relationship to relevant constructs and concepts.”

Interested organizations are invited to submit a two-page “discovery paper” outlining the value proposition, operational impact and end-user demand for their proposed solutions.

After papers are evaluated, vendors may be invited to a “digital proving ground” where they can pitch their innovations to contracting officers who will be looking to make rapid pilot project awards.

The DOD may decide to award other transaction agreements for prototypes to a single vendor or multiple vendors, with the potential for follow-on production awards, according to the announcement.

The end date for the open call is July 3.

The post DOD issues open call to industry seeking new tools for evaluating warfighter trust in AI-enabled systems appeared first on DefenseScoop.

The “Acqbot” prototype is part of the CDAO’s Tradewind initiative.

“Speed to contract in my functional lane is extremely important. And I think it’s going to be a differentiator in whatever cyber wars are going to happen in the next couple [or] few years,” Bonnie Evangelista, Tradewind execution lead, said at the ATARC federal IT modernization summit on Wednesday. “We’re working on a prototype, it’s an AI-powered contract writing capability. So we have a working prototype … We want to see if we can break the mold and break the glass ceiling from an acquisition perspective.”

The aim is to accelerate workflows to help get emerging technology into the hands of service members faster.

“We’re asking questions like, can an AI or can a bot write a contract? Like, is that possible?” she said. “Can vendors write proposals on their phones? … Those are like big, bold statements and those are the things that like from a practitioners’ perspective and contracting — those are the things we’re considering.”

The prototype can help generate text, similar to the way ChatGPT operates, according to Evangelista. However, as the Defense Department experiments with the technology for contract writing, there’s still a person in the loop.

“There’s a human reviewing and validating the text at every point in the process. But the tool is going to help generate certain text to help inform your problem statement. And then that’s just the beginning of the workflow,” she explained. “We have the workflow going all the way from problem statement to helping to generate what we call ‘the call to industry.’” The hope is that industry can also “respond in that workflow” and the AI can generate the contract.

The CDAO is working to train the model to create problem statements based on descriptive inputs.

“I don’t think it’s gonna take a massive amount of data to nuance or train the model. But I will say that we don’t expect, you know, we don’t expect glory overnight,” Evangelista said. “This is something where we’re hoping if we can continue to prove the concept and then create a demand or a demand signal for this type of tool, that’s where we can make it optimize the technology … [and] we’ve given it a chance to be enduring.”

The post AI bot developed to help Defense Department write contracts faster appeared first on DefenseScoop.

With its partners in the Army Contracting Command-Rock Island and the Indiana Innovation Institute (IN3), the CDAO is targeting the first quarter of fiscal 2023 to “go-live” with the minimum viable product of this new “Tradewind Solutions Marketplace.”

Between now and Sept. 30, officials involved are crowdsourcing suggestions from industry, academia and government agencies on the concept and framework underpinning that hub and how such organizations could help “shape” it.

After that date, some comments received may be shared publicly. But those behind the emerging vendor space also plan to continue to collect feedback and engage interested parties throughout the existence of the Marketplace initiative.

In early 2021, DOD’s former Joint Artificial Intelligence Center (JAIC) — which was recently absorbed into the CDAO — and the Army awarded an Other Transaction agreement to IN3 to create a business process and online environment called “Tradewind,” that would drive more efficient AI acquisitions for the U.S. military. Last month, FedScoop reported that the partners recently updated the tradewindai.com website and it is being tested as a modern channel to announce new AI-aligned challenge competitions.

On Aug. 26, the Tradewind Solutions Marketplace announcement was posted there and on the SAM.gov federal government contracting site. Both link to the Tradewind Exchange Challenge Summary landing page where responses are to be submitted.

Specifically, officials want feedback on how they might best provide a venue where defense and military insiders can search for the technologies of interest, and a single location to interact with external organizations that can deliver them through an established rapid contracting pathway.

A 10-page, attached “teaser” draft of the marketplace open call also provides details behind the evolving, and subject to change, notions informing the making of this new online shop — as well as some information about how it will operate.

“The Solutions Marketplace serves industry and academic organizations by providing a forum to showcase relevant research, products, and services to prospective government customers, and serves DOD by providing a forum to access data, analytics, digital and AI/ML solutions and rapidly ingest game changing technology solutions,” the document said.

Envisioned as “a digital environment of competed video pitches,” the marketplace will be designed as a venue for customer organizations “to search, view, review, compare, contrast, contact, negotiate, and procure data, analytics, digital and AI/ML” technologies.

The overarching idea is that once video pitches of capabilities pass through a deep assessment to ensure compliance with federal requirements, and they are approved for the marketplace, they will then be made available for funding via Other Transaction agreements or procurement contracts. 

“Thus, the Tradewind Solutions Marketplace serves as the DOD’s digital environment of post-competition, readily awardable, technology solutions,” the document said.

Video solution pitches will have to address one or more topics on a list of strategic focus areas that will also change over time based on the Pentagon’s needs.

Initially, those areas are: improving situational awareness and decision-making, increasing safety of operating equipment, implementing predictive maintenance and supply, streamlining business processes, assuring cybersecurity and discovering Blue Sky technology applications.

The latter essentially refers to future-facing domains where “real-world” applications are not immediately apparent. The CDAO’s press office did not provide further information by FedScoop’s deadline about what capabilities those involved want in that case.

The post DOD to roll out new online marketplace to speedily buy AI-aligned tech appeared first on DefenseScoop.

FedScoop confirmed this week that the Pentagon’s Chief Digital and Artificial Intelligence Office (CDAO) recently updated its tradewindai.com website, and is now trying it out as a one-stop-shop to announce new AI-aligned challenge competitions and opportunities supporting DOD components. 

“While this by itself is not a new approach, the creation of time-based events and near real-time awards is a new approach for us,” CDAO spokesperson Kathleen Clark told FedScoop on Friday.

On Aug. 16-17, the CDAO will use an industry day event happening near Quantico, Virginia, hosted by the 5th Special Forces Group, to meet with potential commercial partners about several of its open challenges. During those engagements, according to Clark, the CDAO intends “to make an award determination for pilots in areas such as digital STEM kits for recruiting of future acquisition professionals,” and a Southwest-focused regional pipeline to promote the quick transition of college students from minority institutions to positions at DOD test centers.  

“What makes this approach new is that, instead of back-and-forth papers and written interactions, we will leverage real-time, in-person and virtual interactions to allow both industry and the government to better understand the requirements and the art of the possible,” she said.

The ‘edge of possibility’

In early 2021, the Pentagon’s Joint Artificial Intelligence Center (JAIC) and the Army Contracting Command-Rock Island awarded an Other Transaction agreement to Indiana Innovation Institute (IN3) to generate a business process and online environment, called “Tradewind,” that could enable more efficient AI procurement for the U.S. military. Since then, the JAIC has been absorbed into the newly formed Chief Digital and Artificial Intelligence Office — and the Pentagon’s AI procurement work associated with Tradewind is maturing with experimentation playing an expanding role.

Tradewind is — as it was envisioned to be — a continually changing environment, in Clark’s view. It is a prototype initiative, but it’s also more than one single agreement.  

“It is a way for the CDAO to look at new acquisition methods, processes, ideas, or innovations,” she said.

With a sharp, near-term focus on trying new approaches, the office is now experimenting more than before in terms of options around AI purchasing. For instance, officials are exploring how a procurement model that was initially developed to support cyber procurements can be adapted to support XVIII Airborne Corps readiness requirements. 

The office also recently revamped its public-facing portal by launching its new Tradewind website. Clark noted that digital hub had been in Beta mode until Aug. 8. The homepage bills Tradewind as “DOD’s AI Marketplace to find, fund, and accelerate the adoption and transition of AI technologies to solve DOD challenges and problems.”

It now includes links to recruiting and partnership opportunities with DOD components, such as an open post from the Defense Threat Reduction Agency about developing an environment for AI technologies associated with its chemical and biological defense operations.

Clark noted that IN3 “has been and still is the government’s lead partner on Tradewind.” In 2023, that partnership will continue and the CDAO will also work with “many others to perform more sprints, new innovations and then scale them,” she added. Clark confirmed on Saturday that Tradewindai.com is managed by several entities including IN3.

“The entities that facilitate the site do not compete on efforts listed on the site. These are two separate and distinct functions. The government determines challenges and evaluates challenges,” she said.

Compared to the last version, the new site has an adjusted layout to display more pictures and links to other backend databases, so the CDAO can perform additional analyses on interactions with industry and academia around acquisition activities, according to Clark.

“This includes statistics on who is accessing, are they interested, have they submitted a response, has the government reviewed the response, and what is the next step for the government,” she said. “This is all a part of automating our acquisition process and increasing the interactions with industry and understanding what is working and what is not.”

As the CDAO is data- and AI-focused, its acquisitions unit is “taking this concept to the core,” she added — particularly by moving the website from a flat capability to one that connects to data platforms that provide insights on how to improve communication and processes based on industry engagement.

The CDAO’s procurement officials are also pursuing AI-enabled acquisition platforms that can help them write better problem statements and improve their reviews of industry proposals. The automation of acquisition processes to reduce human workload and improve workflows is also of interest.  

“We are just starting to touch the edge of the possible for acquisition, but this shifts DOD’s approach to acquisition from one of risk avoidance to one of agile experimentation and scaling,” Clark said.  

At Tradewind’s inception, JAIC insiders were primarily concentrating on better grasping and interacting with outside companies.

“Now that we have a better understanding of the market we are progressing to agile procurement pilots. This is a natural transition,” Clark said. “Now that we better understand the industrial base and some of the areas that can leverage DOD acquisition innovations, we create sprints to pilot new approaches.”

She likened this approach to software-related sprint cycles that allow teams to try, learn and scale the things that work quickly. 

“During our August 16 and 17 event, we will unveil some of our 2023 initiatives and pilots for acquisition to see what areas industry has the most interest in, as well as partners from Special Operations Command,” Clark added.

Editor’s note: This article was updated with new details about the maintenance of the Tradewind website.

The post DOD trying new approach for Tradewind AI initiative appeared first on DefenseScoop.