The number of chief artificial intelligence officers (CAIOs) has almost tripled in the last 5 years, according to LinkedIn. Companies across industries are realizing the need to integrate artificial intelligence (AI) into their core strategies from the top to avoid falling behind. These AI leaders are responsible for developing a blueprint for AI adoption and oversight both in companies and the federal government.

Following a recent executive order by the Biden administration and a meteoric rise in AI adoption across sectors, the Office of Management and Budget (OMB) released a memo on how federal agencies can seize AI’s opportunities while managing its risks.

Many federal agencies are appointing CAIOs to oversee AI use within their domains, promote responsible AI innovation and address risks associated with AI, including generative AI (gen AI), by considering its impact on citizens. But, how will these CAIOs balance regulatory measures and innovation? How will they cultivate trust?

Three IBM leaders offer their insights on the significant opportunities and challenges facing new CAIOs in their first 90 days:

1. “Consider safety, inclusivity, trustworthiness and governance from the beginning.”

—Kush Varshney, IBM Fellow

The first 90 days as chief AI officer will be intense and speed by, but you should nevertheless slow down not take shortcuts. Consider safety, inclusivity, trustworthiness, and governance from the beginning rather than as considerations to be tacked on to the end. But do not allow the caution and critical perspective of your inner social change agent to extinguish the optimism of your inner technologist. Remember that just because AI is here now, your agency is not absolved of its existing responsibilities to the people. Consider the most vulnerable among us, when specifying the problem, understanding the data, and evaluating the solution.

Don’t be afraid to reframe fairness from simply divvying up limited resources in some equitable fashion to figuring out how you can care for the neediest. Don’t be afraid to reframe accountability from simply conforming to regulations to stewarding the technology. Don’t be afraid to reframe transparency from simply documenting the choices made after the fact to seeking public input beforehand.

Just like urban planning, AI is infrastructure. Choices made now can affect generations into the future. Be guided by the seventh generation principle, but do not succumb to long term existential risk arguments at the expense of clear and present harms. Keep an eye on harms we’ve encountered over several years through traditional machine learning modeling, and also on new and amplified harms we’re seeing through pre-trained foundation models. Choose smaller models whose cost and behavior may be governed. Pilot and innovate with a portfolio of projects; reuse and harden solutions to common patterns that emerge; and only then deliver at scale through a multi-model platform approach.

2. “Create trustworthy AI development.”

—Christina Montgomery, IBM Vice President and Chief Privacy and Trust Officer

To drive efficiency and innovation and to build trust, all CAIOs should begin by implementing an AI governance program to help address the ethical, social and technical issues central to creating trustworthy AI development and deployment.

In the first 90 days, start by conducting an organizational maturity assessment of your agency’s baseline. Review frameworks and assessment tools so you have a clear indication of any strengths and weaknesses that will impact your ability to implement AI tools and help with associated risks. This process can help you identify a problem or opportunity that an AI solution can address.

Beyond technical requirements, you will also need to document and articulate agency-wide ethics and values regarding the creation and use of AI, which will inform your decisions about risk. These guidelines should address issues such as data privacy, bias, transparency, accountability and safety.

IBM has developed trust and transparency principles and an “Ethics by Design” playbook that can help you and your team to operationalize those principles. As a part of this process, establish accountability and oversight mechanisms to ensure that the AI system is used responsibly and ethically. This includes establishing clear lines of accountability and oversight, as well as monitoring and auditing processes to ensure compliance with ethical guidelines.

Next, you should begin to adapt your agency’s existing governance structures to support AI. Quality AI requires quality data. Many of your existing programs and practices — such as third-party risk management, procurement, enterprise architecture, legal, privacy, and information security — will already overlap to create efficiency and leverage the full power of your agency teams.

 The December 1, 2024 deadline to incorporate the minimum risk management practices to safety-impacting and rights-impacting AI, or else stop using the AI until compliance is achieved, will come around quicker than you think. In your first 90 days on the job, take advantage of automated tools to streamline the process and turn to trusted partners, like IBM, to help implement the strategies you’ll need to create responsible AI solutions.

3. “Establish an enterprise-wide approach.”

—Terry Halvorsen, IBM Vice President, Federal Client Development

For over a decade, IBM has been working with U.S. federal agencies to help them develop AI. The technology has enabled important advancements for many federal agencies in operational efficiency, productivity and decision making. For example, AI has helped the Internal Revenue Service (IRS) speed up the processing of paper tax returns (and the delivery of tax refunds to citizens), the Department of Veterans Affairs (VA) decrease the time it takes to process veteran’s claims, and the Navy’s Fleet Forces Command better plan and balance food supplies while also reducing related supply chain risks. 

IBM has also long acknowledged the potential risks of AI adoption, and advocated for strong governance and for AI that is transparent, explainable, robust, fair, and secure. To help mitigate risks, simplify implementation, and take advantage of opportunity, all newly appointed CAIOs should establish an enterprise-wide approach to data and a governance framework for AI adoption. Data accessibility, data volume, and data complexity are all areas that must be understood and addressed. ‘Enterprise-wide’ suggests that the development and deployment of AI and data governance be brought out of traditional agency organizational silos. Involve stakeholders from across your agency, as well as any industry partners. Measure your results and learn as you go – both from your agency’s efforts and those of your peers across government.

And finally, the old adage ‘begin with the end in mind’ is as true today as ever. IBM recommends that CAIOs encourage following a use-case driven approach to AI – which means identifying the targeted outcomes and experiences you hope to create and backing the specific AI technologies you’ll use (generative AI, traditional AI, etc.) from there.

CAIOs leading by example

Public leadership can set the tone for AI adoption across all sectors. The creation of the CAIO position plays a critical role in the future of AI, allowing our government to model a responsible approach to AI adoption across business, government and industry.

IBM has developed tools and strategies to help agencies adopt AI efficiently and responsibly in various environments. We’re ready to support these new CAIOs as they begin to build ethical and responsible AI implementations within their agencies.

Are you wondering what to prioritize in your AI journey?