The 6 Blind Spots That Could Derail Your GenAI Ambitions

Forget the hype. GenAI isn’t just a buzzword; it’s a revolution poised to redefine how we work, play, communicate, and learn.

CEOs are catching on. They see GenAI’s potential to conquer data mountains, crack complex codes, and uncover hidden correlations while freeing us from the mundane and unleashing a torrent of business creativity.

But IT leaders? They’re teetering on the edge. GenAI’s potential is undeniable, but a minefield of concerns holds them back. A recent IBM Institute of Business Value (IBV) study of 2,500 C-suite tech leaders exposed six critical blind spots that could sabotage any GenAI initiative.

The real culprit isn’t the technology itself but organizational readiness. Address these blind spots now, or prepare for a collision course with misalignment and budget overruns.

Reimagine the role of IT

The first blind spot? An outdated view of IT’s role. In today’s hyper-accelerated tech landscape, IT can’t afford to be a mere toolbox provider. This is a non-starter for GenAI, which thrives on augmenting human ingenuity.

To unlock its true potential, IT leaders must position technology at the heart of the enterprise. CEOs get it. The IBV study reveals that product and service innovation top their agenda for the next three years. Yet, a mere 43% of tech leaders believe they’re up to the task. More than half doubt their ability to deliver.

“IT leaders must seize the wheel and guide the business in identifying real-world applications for GenAI,” asserts Gaurav Chhiber, IBM’s vice president for data, AI, and security in Asia Pacific.

Chhiber’s mandate for IT leaders: become GenAI evangelists. Seek out real-world challenges ripe for AI disruption, cultivate a culture of experimentation and collaboration, and establish technology as the innovation engine. It’s not a comfort zone but essential for GenAI’s success.

Break down the walls

Another roadblock: superficial collaboration. The IBV study exposes a chasm in interdepartmental data collaboration, especially between IT and finance. Silos, turf wars, and data ownership anxieties stifle the data sharing essential for GenAI to flourish.

The solution? IT leaders must become master diplomats, weaving themselves into the fabric of enterprise decision-making. Data and analytics are their weapons of influence.

“IT leaders must reimagine their infrastructure and data platforms to fuel GenAI initiatives,” insists Chhiber. “Aligning goals and metrics with finance is crucial for a unified strategy.”

Chhiber recommends leveraging ITFM tools to demonstrate how technology can revolutionize financial planning and management within IT projects. This deep collaboration is GenAI’s lifeblood. With the potential for runaway costs, IT leaders must partner with finance to rationalize investments and align GenAI initiatives with core business objectives.

Join Gaurav Chhiber and other IBM leaders at the Dec. 3, 2024, virtual event "Is Your Business AI-Ready? Or About to Be Disrupted?". To learn more or register for the event, click here. 

GenAI mistakes can break your company

Chhiber warns, “Choosing the wrong AI foundation model (or LLM) can backfire — costing your business money, compromising efficiency, generating inaccurate outputs, and eroding customer trust.” This is another hidden danger. Even with a robust infrastructure, the wrong model choice can derail your GenAI ambitions.

Not every business problem demands a large language model (LLM). While versatile, LLMs can be costly, complex, and unpredictable. IT leaders should guide their companies toward smaller, specialized models.

IBM’s Granite 3.0 LLM, for example, offers a compelling alternative. Customizable, efficient, and cost-effective, it’s a perfect fit for diverse enterprise needs.

“For instance, potential inferencing costs for an enterprise using IBM’s Granite 13B model could be 42x less than using OpenAI’s GPT4,” says Chhiber. “That represents a savings of USD 41,000 per month — money that can be reinvested into the business.”

Importantly, you don’t have to sacrifice accuracy for cost-effectiveness. Model flexibility ensures the choice to select the highest level of performance by the task at the lowest cost and, ultimately, a higher return on investment.

Rein in GenAI irresponsibility

A critical issue — and a prominent blind spot — is trust in AI. As AI becomes increasingly integrated into our lives, it’s crucial to address trust and related ethical implications. Tech leaders must ensure that AI systems are not only effective but also responsible and trustworthy.

It’s not surprising that nearly three in four CEOs (71%) say establishing and maintaining customer trust will have a greater impact on their companies’ success than any specific product or service. And for 80% of CEOs, transparency in their companies’ use of next-generation technologies such as GenAI is critical for fostering that trust.

Chhiber offers five steps to address this blind spot: establishing clear ethical frameworks, prioritizing explainability, implementing human oversight mechanisms, regularly auditing and monitoring AI systems, and fostering a culture of ethics and responsibility.

Three announcements from the recent IBM TechXchange event in Las Vegas reinforce these steps. First, IBM announced IBM watsonx.governance Guardrails, which supports real-time detection. It also unveiled Evaluation Studio, an upcoming feature release of IBM watsonx.governance, which reduces governance time and effort. Lastly, IBM is integrating watsonx.governance with Guardium AI Security to discover shadow AI and address security risks.

Transforming data from liability to asset

It is now common industry knowledge that data is the fuel to power GenAI. However, the reality is that data remains a potential liability rather than a valuable asset for many companies. “To take advantage of GenAI, you need to break down your enterprise silos to connect your data,” says Chhiber.

This is not a new blind spot but one that has existed for a long time. There’s been a lot of effort to consolidate data, but as the IBV report suggested, “enterprise data may appear integrated on a screen, but beneath the surface, collection and integration are often cobbled together manually.”

Even if the data is accessible, tech leaders think they are not ready. In the report, less than a third (29%) strongly agree their enterprise data meets the quality, accessibility, and security standards that support the efficient scaling of generative AI. 45% said that their angst about data accuracy or bias has increased in the last six months because of GenAI.

Chhiber suggests a hybrid-by-design architecture to align and manage large amounts of data on-premises, multicloud, and at the edge. “And you need a data foundation built on open-source and integrated technologies to store, unify, and share across your organization to help you discover insights faster,” he adds.

IBM is also simplifying the hybrid-by-design approach with several announcements at the Las Vegas TechXchange event. Chhiber points to the new data ingestion and orchestration with the integration of dbt and Airflow with watsonx.data’s Presto and Spark engine and a common policy gateway that allows companies to obtain external data governance policies and enforce those policies in watsonx.data. On the data integration front, IBM recently acquired StreamSets, enabling real-time streaming data pipelines.

Attracting and retaining top talent

The most obvious yet most difficult blind spot to overcome is talent. So, when two-thirds of CEOs said their teams had the knowledge and skills to incorporate new tech, such as GenAI, only half of tech leaders shared this optimism in the IBV report. This is not just a GenAI problem; it’s an IT-wide problem that vendors and customers share.

At IBM, Chhiber notes IBM is tackling this concern internally by providing various training opportunities, including self-paced courses, live virtual classes and sandbox environments like the bi-annual watsonx challenge.

The company is rolling out new initiatives to help companies ease their talent gap. At the Las Vegas TechXchange, IBM announced AI Agents for HR, a new generative AI assistant that transforms how HR professionals and employees engage. It paves the way to what Chhiber calls an agentic future.

The GenAI future belongs to the bold

The GenAI revolution is not coming — it’s already here. While the potential benefits are immense, so are the challenges.

Overcoming the blind spots can be an enormous task for a single company. Chhiber advises companies to look for partners who are invested in their success.

“At IBM, for example, we have invested in a 100+ strong Client Engineering team across APAC to co-create with our clients to identify use cases and take these to pilots,” says Chibber. “We have seen that this has greatly accelerated the speed of innovation with many of our clients.”

The future belongs to those bold enough to embrace the transformative power of GenAI and wise enough to navigate its complexities with foresight and integrity.

Image credit: iStockphoto/Jake Lomachevsky