On December 5, 2023, IBM and Meta launched the AI Alliance, with partners from around the world, to challenge proprietary models and opaque practices that will likely lead to distrust of AI over time. I joined an analyst call this morning to hear from principals from founding members. The idea is to drive a vibrant and inclusive, open community of innovators. The initial membership list draws from industry, startups, academia, research and government organizations.
IBM Chairman and CEO Arvind Krishna shared the following in the analyst briefing documents: “The progress we continue to witness in AI is a testament to open innovation and collaboration across communities of creators, scientists, academics and business leaders. This is a pivotal moment in defining the future of AI. IBM is proud to partner with like-minded organizations through the AI Alliance to ensure this open ecosystem drives an innovative AI agenda underpinned by safety, accountability and scientific rigor.”
Some of the over 50 initial members include AMD, Anyscale, CERN, Cerebras, Cleveland Clinic, Cornell University, Dartmouth, Dell Technologies, EPFL, ETH, Hugging Face, Imperial College London, Intel, INSAIT, Linux Foundation, MLCommons, MOC Alliance operated by Boston University and Harvard University, NASA, NSF, Oracle, Partnership on AI, Red Hat, Roadzen, ServiceNow, Sony Group, Stability AI, University of California Berkeley, University of Illinois, University of Notre Dame, The University of Tokyo, and Yale University,
The Alliance will focus on benchmarks and evaluation standards, advancing the ecosystems responsibly, driving hardware innovation and access, global AI skill building, and education.
From its FAQ on Why is Open Innovation Essential to AI:
- Democratizes access to the most foundational and broadly applicable advances;
- Harnesses the innovative talent of the global community;
- Ensures accountability among individuals and companies;
- Instills trust from transparency by demystifying technical innovation for both the public and policymakers and
- Better enables robust testing and validation through broad-based community approaches.
Nick Clegg, President of Global Affairs at Meta, emphasizes some of those key points in his statement: “We believe it’s better when AI is developed openly—more people can access the benefits, build innovative products and work on safety. The AI Alliance brings together researchers, developers and companies to share tools and knowledge that can help us all make progress whether models are shared openly or not. We’re looking forward to working with partners to advance the state-of-the-art in AI and help everyone build responsibly.”
AI Alliance: Analysis
The AI Alliance lists an impressive roster of members. Not among them, however, is OpenAI, Microsoft or Google. Nor does it include Salesforce or Cisco. That is not a surprise at this point. The proprietary and profit-seeking investments from the leaders in the market are what is driving the formation of this alliance. Meta’s inclusion is a win for IBM, which has increasingly moved to open source over the last several years. While plenty of proprietary software still serves up value to IBM and its customers, many well-known proprietary systems, such as Lotus Notes, have disappeared from their catalog. IBM is increasingly the implementor of open-source software.
The tension between open-source and proprietary software is decades old. AI is just the latest technology to fall into the open-source versus proprietary spectrum.
Proprietary AI may lead to breakthroughs and economic models that drive investments, serve profitable niche markets, and return value to shareholders. But it risks doing so in ways that may prove harmful to individuals or to society at large.
The open-source community would argue that proprietary solutions are not the only path. Open source will offer more transparency and a word often missing from responsible AI policies: accountability.
What this means in the near term is competition, which has been restricted to the big players despite the plethora of models on Hugging Face. People are playing with many models and commercializing a few. The integration of AI across Microsoft’s offerings and its integration with Google Workspace creates easy access to their tools, limiting exposure to alternatives. Even if their development tools offer agnostic access, implementations will likely run on the platforms of record for the cloud service providers upon which solutions are built.
Because of the AI Alliance’s broad remit, they will be better positioned to help general organizations manage issues like duplication of function that will likely start to appear in 2024 as organizations adopt more than one AI solution within the same functional area.
What they may not be able to do is manage expectations for existing systems, like ChatGPT or Bard, that may prove difficult to retrofit with any of the more open technologies created by the alliance. It is highly unlikely that existing systems will be rebuilt under new rules and retain their identities. In the near term, systems like ChatGPT and Bard will have rules and other interventions bolted onto them, which will be workable if not completely effective. That approach will eventually become untenable. The generative AI tools we know today will be the first legacy generative AI applications.
At some point, the initial generative AI systems will need to be retrained and rebuilt on new technology that incorporates many of the features the AI Alliance will eventually advocate. That, or they will be rebuilt, taking only “useful” technology from the Alliance but remaining opaque on many of their other internal choices.
The AI Alliance points toward the flaws in commercial models that cast doubt, even fear, in workers and consumers. If people know more, and the systems permit more visibility into their workings, then perhaps they will be more comfortable. That is a good goal, but one that does not reflect the history of education or good intentions.
Because the future is uncertain, I encourage the AI Alliance to engage in scenario planning early and to maintain a set of Future of AI scenarios as the factors that cause fear, reluctance, and also opportunity that comes not just from technology but from social and political arenas as well. A group like the AI Alliance should be the home to the large canvas upon which the future of AI is monitored, not just the seat of technologies and approaches that live in single futures or, more negatively, across multiple, unarticulated futures that will make it difficult for people to reconcile their beliefs. Scenarios provide a visible workshop for making possibilities explicit as well as exploring them.
The AI Alliance certainly offers an alternative vision to proprietary AI development, one that will likely sit well with governments and consumers. Alliances, or trade associations, have been around for a long time, often driving innovation in the background while negotiating common objectives. Computer users will know the work of trade associations via their specifications, such as Bluetooth® and USB.
While all development requires negotiation, trade associations require much broader negotiation with partners. Unlike internal organizations, they tend to act as meritocracies rather than hierarchies. The point is that management models may continue to offer leverage to commercial organizations that can decide on issues and invest faster. Trade associations, however, often have access to a broader array of talent who may be able to offer a wider variety of options for overcoming obstacles and more inclusive solutions, given a diverse talent pool.
How well-managed the AI Alliance will prove to be remains to be seen. Its ability to define goals and deliver on those goals will be critical, as will its ability to track developments and perhaps anticipate them so that it remains nimble as the technology and market changes. The fast pace of AI development will likely prove to be a management challenge to the AI Alliance as it attempts to keep pace with the fast-evolving technology in history.
Given the breadth and reputation of early membership, those not participating will be watching for the association’s efficacy. Those who are members need to invest time and talent to ensure that the AI Alliance has the resources it needs to succeed. Underinvestment by members will likely lead to an ineffective AI Alliance.
For more serious insights on artificial intelligence, click here.
Leave a Reply