Pursuing AI Responsibly: A Business Technologist Perspective

While it may be challenging for near-term generative AI (GenAI) use
cases to live up to the hype of the past two years, the consensus is that the technology is
a potential game-changer for business innovation, creativity, personalization,
efficiency and new business models. It is also being touted — somewhat
ironically, given its intensive use of resources — as a solution to many
challenges related to climate change and sustainability.

AI’s risks and
challenges
— ethical, social and environmental — have been left out of the business hype
for these tools. The risks have recently gained the attention of regulators and
global institutions including the European Commission, White House and United
Nations; but they’re still not front and center among business leadership and
typical AI users. This has generated alarm among information technology (IT)
executives, who are sensing a pattern with which they are all too familiar.

Like many widely hyped, new technologies (cloud computing, IoT, blockchain and
robotic-process automation come to mind) GenAI has been hastily adopted — often
without an enterprise strategy, business case or governing ruleset. According to
the “Responsible AI” section of Info-Tech’s Tech Trends
2024 report, 35 percent
of surveyed companies deploying AI lacked formal AI governance guidelines, and
less than a third conducted AI impact assessments. Such shortcomings typically
lead to disappointing results, costly or embarrassing misuses of the technology,
and unacceptable risk.

It has typically been up to IT organizations to bring order to innovation chaos
— ensuring that technology scales efficiently, securely, with proper
integration, and ongoing monitoring. But compared to previous technology waves
such as the cloud, the stakes for AI are higher. Among the business risks of
poorly governed
AI
are the following:

Undermined ethics: The use of GenAI’s creative abilities raises rights
and ownership dilemmas for intellectual property and data. The misuse of
GenAI, for deep fakes and other forms of
misinformation,
can cause significant business and societal harm.
Exacerbating the digital divide: Without broad inclusivity and equitable
access to AI applications and their benefits, the technology could widen
society’s digital divide.
Loss of integrity: AI decisions are susceptible to inaccuracies and
discriminatory outcomes due to biases in data and prompts — aka
“hallucinations.”
Low-quality or biased outputs can harm business reputations and erode
customer and institutional trust.
Resource intensity: The energy and resource use by the AI industry,
including chip manufacturing and data center facilities, is staggeringly
high.
- Since 2012, the most extensive AI training runs have been using
  exponentially more computing power, doubling every 3.4
  months, on average.
- OpenAI’s GPT-3 training is estimated to have used 1.3
  gigawatt-hours of energy (equivalent to 120 average US households’
  yearly consumption) and generated 552 tons in carbon emissions
  (equivalent to the yearly emissions of 120 US cars).
- A user’s prompt or query on ChatGPT uses 10x the power
  of an equivalent traditional Google search.
Other societal risks: AI competes with agriculture and municipalities
for finite energy and water resources. The recent industry push for nuclear
power comes with its own
long-term social and environmental risks:
- Job displacement: While AI can augment human work, there is also
  concern about large-scale job displacement — particularly in industries
  where routine tasks are easily automated.
- Security and privacy breaches: AI can be exploited for malicious
  purposes, such as creating fake identities or generating harmful
  content. Use of data to train and inform AI algorithms may inadvertently
  violate privacy laws or regulations.

The IT executive members of SustainableIT.org
— a non-profit professional association dedicated to driving sustainability
through technology — want to help businesses avoid these negative impacts while
maximizing the transformational benefits of their AI deployment. To that end, in
September, they developed and published a framework and set of principles to
guide responsible AI application deployment. While the guidance is uniquely
informed by the business technologist perspective, it is pertinent to leadership
from the boardroom to the C-suite in every industry, and our goal is that it
will be used for education.

The framework offers a simple yet comprehensive model that incorporates three
stages and nine principles:

Stage 1 – Reflect

Consider intended uses and desired outcomes of AI applications — assessing
potential positive and negative impacts to business stakeholders, strategies,
goals and commitments.

Related principles:

Risk due diligence: AI applications are thoroughly analyzed before
deployment at scale for their risk materiality to — and implications for —
business operations, policies, compliance, goals and strategies.
Sustainability due diligence: AI applications are thoroughly analyzed
before deployment at scale for current and long-term implications for ESG
commitments, policies and
regulations.
Ethical usage: AI application deployment and use are monitored for
alignment to and compliance with the organization’s ethical standards and
business values (e.g., equity and inclusion, nondiscrimination,
transparency, safety).

Stage 2 – Reframe

Redevise governance rules, processes, roles and skill sets — as well as
enterprise operations and architecture — to maximize AI benefits and avoid
negative impacts.

Related principles:

Data optimization: Data used in AI applications are appropriate,
transparent, secure, privacy compliant, consensual and as unbiased as
possible — supported by appropriate data governance.
Trustworthy outcomes: Results, recommendations and decisions made or
informed by AI applications are fair, reasonable, explainable, accurate and
cause no harm to human health, safety or fundamental rights.
AI literacy: AI application deployment coincides with development of
users’ understanding of AI operations, limitations and risks; and the
knowledge to apply AI appropriately and effectively in their roles.

Stage 3 – Reimagine

Conceptualize new business applications, processes and experiences uniquely
suited to AI’s ability to augment human capabilities through collaboration and
automation.

Related principles:

Human first: AI deployment prioritizes enhancement and augmentation of
existing jobs/roles, with upskilling and prioritized redeployment of
displaced workers.
Inclusive benefits: The benefits of AI are equitably applied to,
accessible and leveraged by the broadest range of targeted stakeholders; and
do not intentionally or unintentionally exclude disadvantaged groups.
Responsible innovation: AI-dependent innovation goals and outcomes are
subject to the same governing criteria for risk, sustainability and ethics
that are applied to AI usage.

Join the change

This Responsible AI
Framework
is only a preliminary step. SustainableIT.org has formed a Responsible AI
Working Group — open to all interested organizations — with the mission to
inform and equip organizations worldwide with guidance and tools to govern the
efficient, secure and sustainable implementation of AI. The group will research,
curate and adopt the best existing tools and create new resources to fill gaps
in areas including AI literacy; data confidentiality, integrity and
accessibility; and AI cost-benefit models. Its output will be shared with global
businesses, institutions and executives from all business functions; and will be
provided to the United Nations to inform the Global Digital
Compact and AI For
Good initiatives.

It is imperative to establish or elevate responsible governance for GenAI, and
organizations should turn to their IT leaders to drive it. Then, AI may indeed
live up to its hype — transcending the boundaries of traditional computing speed
and complexity to help humans create outcomes barely imaginable today.

Rick Pastore

Research Principal, SustainableIT.org

Rich Pastore develops and delivers research-based insights, tools, case histories, and other content tailored for IT leaders driving ESG strategies for their functions, enterprises and industries. He has over 25 years of experience working with CIOs and their teams to apply thought leadership and best practices to maximize business value from information technology.

Published Nov 18, 2024 8am EST / 5am PST / 1pm GMT / 2pm CET

Source link