This content is paid for by an advertiser and published by WP Creative Group. The Washington Post newsroom was not involved in the creation of this content. Learn more about WP Creative Group.
Content from IBM

A matter of trust:

Why businesses must ensure generative AI is ethical, accurate and bias-free

A group of people standing on a white floor with the word watsonx.

By WP Creative Group

September 28, 2023

Businesses across all industries are embracing generative AI, and with good reason. This new form of artificial intelligence can help organizations improve key processes like customer service and content creation, support employees with special needs and enable game-changing innovations yet to be dreamt of.

But if it’s not implemented with care, the technology – which can act like it has a mind of its own as it authors new material – can reinforce biases, compromise customer trust and privacy, spread disinformation and ultimately, harm brand reputation. “This poses a huge amount of burden from the user perspective,” said Sean Im, president and CEO at Samsung SDS America, which is using generative AI to automate internal functions like coding and tech support.

So how can businesses benefit from generative AI without falling victim to its downsides? It’s a crucial question as the technology gains steam. By 2032, generative AI is expected to be a $1.3 trillion industry – up from just $40 billion in 2022.*

Establishing “trustworthy AI,” industry shorthand for AI that operates in ways that are ethical, secure and bias free, starts with the data. Enterprises must be able to control the raw information that drives their generative AI apps and services. And they must have tools to verify and optimize that data while filtering out information that could be harmful, illegal or inaccurate.

A woman sitting at a desk with a watsonx logo.

Solid foundations

The good news is that new technology is giving enterprises an alternative to general purpose Large Language Models that essentially scrape the entire web – including its darker corners – to inform their foundation models. A solution like IBM’s watsonx environment, unveiled earlier this year, is purpose-built for enterprise use so organizations can confidently create custom generative AI applications that are trustworthy.

watsonx enables the creation of foundation models that are trained and tuned with carefully curated data, whether from internal or external sources. “There’s always conversations around bias and issues related to the ingestion of data that reflect the realities of the world, which we don’t want to see reflected in the applications of the technology,” said Ryan Hagemann, co-director of IBM’s Policy Lab. “We’re hyper-focused on the need to appropriately curate data sets.”

watsonx provides “a trusted ecosystem,” said Hagemann. Through watsonx, enterprise users can filter, fine tune, train and vet the data that drives their generative AI models, applications and services.

Samsung SDS America is using generative AI to automate everyday processes such as drafting meeting minutes, translating emails, coding and IT support. More ambitiously, it’s using watsonx to add conversational abilities and problem solving to its mobile device lifecycle management system through a platform called Zero Touch Mobility. Longer term, Im sees the technology as having profoundly transformative impacts. “There’s a huge amount of untapped potential, including harnessing massive amounts of data stored in mobile devices in order to drive worker productivity,” said Im.

A plane is parked on an airport tarmac.

watsonx’s ability to create trusted services played a big part in Samsung SDS’s decision to choose it over rival platforms. “Generative AI does not have the ability to discern right or wrong,” said Im. “Not only do we need to look at the right and the wrong of the data, but is it ethical? Does it represent the brand promise? AI is as good as the data you’ve trained it on, and most of that data is generated by humans,” said Im. 

Im believes AI must be transparent, bias free and accountable. “When you ask AI for an answer, and it gives an answer, you want to know how it reached that conclusion.” As the digital arm of consumer electronics giant Samsung, which operates in more than 200 countries, it’s also essential that Samsung SDS’s generative AI services are sensitive to local norms. “Each one of these countries has its own language, customs, ways of working and culture,” noted Im.

Im and his team are hopeful that IBM can solve these issues, given the company’s history and its focus on business users. “IBM has been a leader in data. They know more about data than anyone else. They were the only ones who from the outset that approached this with corporations in mind. What we’re looking for is very, very different from what the general masses might be looking for,” said Im, adding that “watsonx addresses holistically the issues that AI presents from an enterprise perspective.”

An image of a circle with lines in it.
A drawing of a geometric circle.

The right tools

watsonx breaks down the creation of generative AI applications into three buckets, each governed by a specific module:

Such capabilities led Stephanie Soetendal to explore watsonx. Soetendal is founder and CEO at Matrix Holograms, a Boston-based provider of virtual classes designed to make individualized learning more accessible. The company is implementing generative AI to enable students to have live conversations with virtual educators, on laptops or as holograms.

Soetendal said the fact that watsonx allows Matrix Holograms to easily curate trusted data attracted her to the platform. “Due to my experience of over 14 years in the private education sector, I know exactly the textbooks and the sources we need. And that is where generative AI can obtain high-quality material and be respectful of copyrights,” said Soetendal. “We have more control from the beginning on how we’re fine tuning our model.”

Soetendal found watsonx’s tools for data verification through watstonx.governance particularly valuable. “Education is very fragile. If we’re teaching the wrong information, that’s extremely dangerous. It also impacts our reputation. That is why we have set the hallucination of the generative AI, powered by watsonx to zero,” she said.

A group of people pointing at a whiteboard.

Soetendal believes that generative AI can help close the education gap for kids in underserved communities, where schools often lack the latest textbooks, by exposing students to a wider corpus of knowledge. “What we can do through technology is offer all the information on a subject that is available while providing students with critical thinking skills to deduct their own conclusions for an unbiased education.” She views Matrix Holograms’ services as supplemental to the existing educational system. “We’re not replacing teachers.”

Policies for a trusted future

Beyond innovating around generative AI, IBM is advocating for policies that will set a level playing field for responsible use of the technology. “We want to make sure that regulators are taking a scalpel-like approach, as opposed to painting all of its uses with a broader brush,” said IBM’s Hagemann.

In a recent address to technology industry and policy leaders at Majority Leader Charles Schumer’s AI listening forums, IBM CEO Arvind Krishna called for regulators to focus on what he called “high-risk uses of AI.” Among these, Krishna listed propagating misinformation, introducing bias into lending decisions and compromising election integrity. “We strongly believe that regulation must account for the context in which AI is deployed,” Krishna wrote. “IBM clients’ data is their data, and their insights are their insights. We believe that government data policies should be fair and equitable and prioritize openness,” IBM has stated.

This all fits with Big Blue’s larger perspective on AI – that this potentially world-changing innovation must serve humans, not the other way around. “We talk about AI in terms of augmenting human intelligence, and not replacing it. We think that, first and foremost, you need to center individual humans in the implementation and use of the technology,” said Hagemann.

Learn more about how generative AI can be game-changer for your enterprise. 

Learn More

Sources

  1. *Bloomberg