AI, Ethical AI, and Open Source AI
We were able to find a few data points about the level of investment in AI by enterprises in the U.S., but we had to expand to North America and worldwide due to the overall lack of U.S.-specific data. Nonetheless, the data clearly shows that enterprise investment in AI in the past few years has increased substantially. Furthermore, the projections for the level of investment in AI by enterprises in the years ahead are enormous, with approximately six-fold growth expected to transpire between 2020 and 2025.
Enterprise AI Investment
- In 2019, a "survey[ of] 1,000 U.S. business executives at companies that are already investigating or implementing AI" found that "20% said their organizations plan to implement AI enterprise-wide in 2019."
- During Q1 2019 in the U.S., about $2 billion was spent in "funding of AI-focused companies."
- During Q2 2019 in the U.S., there were 139 "AI-related investment deals."
- In 2018, the U.S. had 539 "AI-related investment deals."
- In 2018, a global survey by Deloitte found that 47% of companies were spending between $500,000 and $5 million on AI during that fiscal year, while 34% planned to spend over $5 million, and 17% planned to spend less than $500,000. When asked about their AI spending plans for the following fiscal year, 39% of companies planned to increase spend between 10-20%, 36% of companies planned to increase spend between one and nine percent, 12% planned to increase spend by over 20%, and 11% planned to spend the same amount as they did in FY 2018.
- With regard to future growth estimates, the following are projections published by Statista of expected "[r]evenue from the artificial intelligence for enterprise applications market in North America" between 2020 and 2025: 2020 (approximately $2.28 billion); 2021 (approximately $3.66 billion); 2022 (approximately $5.63 billion); 2023 (approximately $8.23 billion); 2024 (approximately $11.39 billion); and 2025 (approximately $14.89 billion). A screenshot of that data graph is also included in this Google Doc.
- The global revenue projection for enterprise AI by 2025 is estimated at $107.3 billion.
We were only able to find a few data points about the level of enterprise investment in AI specifically within the U.S. between 2015 and 2019. However, we were able to find more information about the level of enterprise investment in AI in North America during that time range. We included data about North America because it was more closely related to the U.S. than global data. In looking for data about the U.S., we implemented three research approaches. First, we reviewed the Statista database, which publishes many reports about market size. There were only a few data points about enterprise investment in AI in the U.S. within Statista and we included each of those findings above. Second, we conducted broad searches for any articles that might have provided that information. Despite numerous searches, the only relevant report we found was from PwC and we cited to data from that report in one finding above. Third, we expanded the scope of our research to areas outside the U.S. In so doing, we found insights pertaining to North America and the level of investment globally, which we provided above.
Companies That Are Vocal About Ethical AI
- Facebook has been vocal about ethical AI.
- Facebook is both talking about the importance of ethical AI and leading the way in developing its own internal frameworks for ethical AI.
- The company summarizes its commitment to ethical AI in stating the following: "At Facebook, ensuring the responsible and thoughtful use of AI is foundational to everything we do — from the data labels we use, to the individual algorithms we build, to the systems they are a part of."
B. Internal Developments
- Facebook has an initiative called Inclusive AI, which is led by Lade Obamehinti. She explained the importance of not just talking about ethical AI, but actually doing something about it in stating, "You can have roundtable after roundtable about this topic without touching the product at all, so don’t get stuck on trying to have this perfect solution or framework from the get-go. It’s really a matter of having concepts, iterating, and trial and error, and that’s what it’s going to take to build lasting frameworks."
- Fairness Flow is a tool that Facebook developed in furtherance of ethical AI, "which it’s using to detect bias in its AI models and has already applied to its jobs algorithm."
C. External Involvement
- In January 2019, Facebook "partner[ed] with the Technical University of Munich (TUM) to support the creation of an independent AI ethics research center . . . [called] The Institute for Ethics in Artificial Intelligence." Facebook provided a funding grant for the Institute in the amount "of $7.5 million over five years." In support of its contribution to the Institute, Facebook said that "AI poses complex problems which industry alone cannot answer, and the independent academic contributions of the Institute will play a crucial role in furthering ethical research on these topics."
- Facebook has plans to continue helping the Institute, including by "offer[ing] an industry perspective on academic research proposals, rendering the latter more actionable and impactful." The company also said it might help the Institute by sharing tools, insights, "and industry expertise related to issues such addressing algorithmic bias, in order to help Institute researchers focus on real-world problems that manifest at scale." As Facebook said, "The Institute is an exciting step forward in our continued commitment to partnering with academic institutions, governments, NGOs, advocacy and industry groups, and others who are working to advance AI in a safe and responsible way."
- Facebook is involved in an initiative called AI4People and the organization Partnership for AI (which Facebook co-founded).
- In December 2019, Facebook sought RFPs for its "Ethics in AI Research Initiative for the Asia Pacific [region]." Facebook sought those proposals because it "is supporting independent academic research work in the Asia Pacific [region] in the field of AI ethics that takes into account different regional perspectives."
- Microsoft has been vocal about ethical AI.
- Microsoft is both talking about the importance of ethical AI and leading the way in developing its own internal frameworks for ethical AI.
- A source described Microsoft as an entity that's "at the forefront of establishing both the ethical guidelines and research and development (R&D) of artificial intelligence."
- Another sourced credited Microsoft with being "one of the earliest companies to begin discussing and advocating for an ethical perspective on artificial intelligence."
B. Internal Developments
- An internal group that Microsoft created for the purpose of promoting ethical AI is called FATE, which stands for "Fairness, Accountability, Transparency, and Ethics in AI."
- Microsoft "has also been working on an internal tool to identify algorithmic bias."
C. External Involvement
- Microsoft announced in July 2019 that it was investing "$1 billion . . . in an OpenAI ethical artificial intelligence project backed by Tesla's Elon Musk and Amazon. The partnership will be devoted to developing advanced AI models on Microsoft's Azure cloud computing platform while adhering to 'shared principles on ethics and trust.'"
- Microsoft established its own school for AI through which people "can learn everything from ethics to building some of the most advanced neural networks in the world."
- AI for Good was established by Microsoft as "an arm of . . . [its] AI R&D." Grants and AI technology donations are provided through AI for Good, "in order to help altruistic individuals and groups bring their ideas to fruition."
- In 2017, Eric Horvitz, the leader of R&D at Microsoft, along with Brad Smith (the Chief Legal Officer and President of Microsoft) "form[ed] Aether, a cross-functional committee addressing AI and ethics in engineering and research."
- Microsoft established, in 2018, "a full-time position in AI policy and ethics." Tim O'Brien a 15-year Microsoft veteran, was the first person to hold that role.
- On its website, IBM has a multi-page section devoted to ethical AI.
- IBM is both talking about the importance of ethical AI and leading the way in developing its own internal frameworks for ethical AI.
- The company developed "new Trust and Transparency capabilities for AI on the IBM Cloud." Those "capabilities provide visibility into how AI is making decisions and give recommendations on how to mitigate any potentially damaging bias. It features a visually clear dashboard that line-of-business users can easily understand, reducing the burden of accountability from data scientists and empowering business users."
- IBM published a guide called "Everyday Ethics" pertaining to ethical AI, which is available on its website.
- IBM created a toolkit called "AI Explainability 360 Open Source Toolkit. " The toolkit helps people "examine, report, and mitigate discrimination and bias in machine learning models throughout the AI application lifecycle." The toolkit includes more than "70 fairness metrics and 10 state-of-the-art bias mitigation algorithms developed by the research community."
- IBM created a position dedicated to ethical AI, which is titled AI Ethics Global Leader. Francesca Rossi holds that title at IBM.
- In January 2020, IBM "issued policy proposals . . . ahead of a . . . panel on AI to be led by Chief Executive Officer Ginni Rometty on the sidelines of the World Economic Forum in Davos. The initiative is designed to find a consensus on rules that may be stricter than what industry alone might produce, but that are less stringent than what governments might impose on their own."
Media Articles About Ethical AI
- The first article we found about ethical AI is titled "EU beats Google to the punch in setting strategy for ethical A.I."
- Elizabeth Schulze is the author of the article.
- CNBC published the article online.
- The article is about guidelines proposed by the European Union with regard to establishing AI ethics. The article noted that those guidelines proposed by the EU were among the first by a governmental entity, as most of the other efforts have come from companies.
- The coverage in the article was both positive and neutral.
- The second article we found about ethical AI is titled "Meet the Researchers Working to Make Sure Artificial Intelligence Is a Force for Good."
- Alejandro De La Garza is the author of the article.
- Time Magazine published the article online.
- The article discusses how academic institutions and employees at tech companies are helping to raise awareness about ethical AI. The article also discusses some of the shortcomings of AI with regard to ethics and challenges that companies are facing and will continue to face, even as they work to embrace ethical AI.
- The coverage in the article was mainly negative and skeptical.
- The third article we found about ethical AI is titled "Ethics And AI: Are We Ready For The Rise Of Artificial Intelligence?"
- The author of the article was not provided, but the article was published online by The Roanoke Star.
- The article discusses how AI employment is outpacing addressing ethical issues inherent in AI. In the article, commonly agreed-upon principles about AI ethics are listed. The lack of AI ethics at U.S. companies is discussed in the article and the article calls out the slow progress being made with regard to ethical AI.
- The coverage in the article was mainly negative.
- The fourth article we found about ethical AI is titled "Real or artificial? Tech titans declare AI ethics concerns."
- Rachel Lerman and Matt O'Brien are the authors of the article.
- AP News published the article online.
- The article discusses that though AI ethics boards are good for companies to have, they alone are not sufficient to ensure that AI is used ethically. The need for government regulation is noted in the article, as is the fact that as companies work to implement ethical AI measures, public scrutiny follows as a byproduct.
- The coverage in the article was mainly negative and skeptical.
- The fifth article we found about ethical AI is titled "Teaching Kids The Ethics Of Artificial Intelligence."
- Cristina Quinn is the author of the article.
- WGBH published the article online.
- The article discusses a summer camp for kids called AI Ethics, which was hosted by Empow Studios and was held at MIT. The AI Ethics camp taught kids about how AI is not perfect. The article also discussed the importance of teaching kids about ethical issues involved in AI, in a way they can understand, so that one day they can be more-educated consumers and professionals.
- The coverage was mainly positive.
Open Source Marketplaces
Since the number of open source marketplaces (including ones focused on AI) in the U.S. was unable to be determined (explained in the Research Strategy section below), we identified three open source AI marketplaces in lieu thereof. The three such marketplaces we identified were Modzy, Shortest Track, and Acumos.
- Modzy is described as an "[e]nterprise AI platform and marketplace offering scalable, secure, and ready-to-deploy AI models." Through the marketplace, people can "view dozens of AI models" that can help them solve complex challenges in the areas of video analysis, natural language processing, and computer vision, among others. The marketplace offers consistent access to API, through which models can be implemented "with just a few lines of code via" Modzy's SDKs and APIs.
- Central management of AI models is also offered through the marketplace. Modzy also offers "enterprise grade security" that enables AI assets to be protected. Auditing and logging is offered, plus Modzy is able to be deployed on private clouds, VMs, bare metal, "or the public cloud."
2. Shortest Track
- Shortest Track is an open source AI marketplace. We included screenshots of the pertinent parts of the company's website in this Google Doc because though the website works, the link to it did not for some reason.
- The marketplace enables users to "create catalogs of solutions in verticals that are currently merchandising." The marketplace also provides that "[a]nalytics management within the enterprise can be accomplished via a private marketplace."
- Through the marketplace, users can "[b]rowse a catalog of millions of open-source algorithms — or choose the right tools for your business from lists evaluated and distilled by our experts." There are filters and search tools within the catalog that help users find the most-beneficial algorithms for their organizations' particular needs. Furthermore, the "catalogs make it easy to organize proprietary and private analytics solutions, keeping each solution secure through logging and access control."
- The marketplace also enables users to "[s]ource top analytics from developers in top laboratories, to create tailored best-of-breed solutions" for the particular challenges that their companies face.
- Acumos is an open source AI marketplace.
- Acumos is described as "an AI marketplace where applications can be chained to create complex and sophisticated AI services."
- The marketplace offers "the capability to edit, integrate, compose, package, train and deploy AI microservices."
- Through the marketplace, users can collaborate, share, and experiment "in an open source ecosystem of people, solutions and ideas."
- The marketplace offers solutions for video analytics, threat prediction, network security, and anomaly detection.
- Acumos also offers a variety of onboarding tools for SciKitLearn, H2O, Java, RCloud, and TensorFlow.
We looked for, but were unable to find the number of open source marketplaces in the U.S., including ones focused on AI. We looked for that information using three research approaches. First, we conducted broad searches for any articles that might have provided that information. The only articles we found mentioned a few such marketplaces, such as this article from Open Source. Second, we checked for that information in the Statista database, but the only information we found was about programmatic spend. Third, we looked for pre-compiled lists of open source marketplaces because had we found any, we might have been able to hand count those to provide an estimated figure. However, we didn't find any lists that mentioned more than just a few such marketplaces, so that method also didn't yield the information we sought. In those research approaches, we also looked to see if information was available at a level outside the U.S., such as for North America or worldwide, but we didn't find any such information for those geographic scopes either. Accordingly, we identified three open source AI marketplaces in lieu thereof.