What IT Leaders Can Learn From Google’s Shaky GenAI Start
Discover why Google Bard’s expansion to platforms like Gmail & YouTube falls short of OpenAI’s response quality buzz.
Peter Pezaris, New Relic’s chief design and strategy officer, compares Google Bard’s AI with OpenAI’s ChatGPT, which explains why prioritizing relevant issues ensures customer benefit, documents thoroughly, and advocates proactive integration strategies for IT teams.
When Google launched Bard, its answer to OpenAI’s ChatGPT, it made an unfortunately timed factual error in its first demo. Google subsequently expanded Bard to integrate with everything from Gmail to YouTube. Still, the public’s early focus was on the tool’s failure to match the quality of OpenAI’s responses.
Google’s false start with generative AI demonstrated the value of accurate outputs from large language models (LLMs). But more importantly, it raised several questions that every IT leader should ask themselves as they look to seize their generative AI opportunity.
To start, learn from our friends in Mountain View and ask yourself the following questions.
What Existing Problem Can We Solve With AI?
People assume that, as a new technology, generative AI must solve new problems. That’s a misconception. Reverse your thinking and ask, “What problems are my team or my company facing, and what would it look like if we addressed those problems with AI?”
Will My AI Solution Truly Help Our Customers?
Just about every tech company is jumping on the generative AI bandwagon. However, only a handful are genuinely adopting and implementing the technology meaningfully that benefits customers.
Many AI chatbots will stick to surface-level use cases and fail to make a real impact. The ones who buck this trend will be the ones who target use cases that were once impossible to solve but now, with generative AI, are merely difficult, like automatically generating a query from an obstructed language.
As your IT team deploys AI solutions across the company, take a high-level look at the outcomes to ensure they align with the company’s goals. The ultimate objective for multi-modal generative AI applications (e.g., mobile, web, voice, chatbots, image, and video) is to move beyond incremental improvements in responses and instead resolve a customer’s issue or answer their question before they even ask it.
Soon, I expect to see proactive generative AI agents that can do things of their own accord because they know they should. This will require AI to have end goals and paths to reach them and a complete understanding of the business.
Is Our Documentation in Order?
Documentation will make or break an AI implementation. You can build the best AI assistant in the world, but if you haven’t documented your business correctly, the assistant won’t understand what your company does and why.
AI cannot guide your employees or customers and help them use your product without that knowledge. Generative AI requires a knowledge graph — a semantic map — of the entire business. Define the nodes and edges as relationships, and the AI assistant will naturally use them in its responses, dramatically increasing the value of generative AI.
Next Steps
- Focus on the future: Many companies view AI primarily as a way to reduce costs; the bigger opportunity lies in innovations that augment human capacity and grow the size of the pie. Focus on new roles the technology will create, as well as the resources, tools, and education the people who fill them will need.
- Resist the temptation: Resist the urge to let generative AI percolate up to the organization on the assumption that innovative ideas only come from below. To drive competitive advantage, organizations need a prescriptive approach. Smart CTOs will proactively support further integration of generative AI into their team’s daily tasks and processes while looking for opportunities to infuse generative AI throughout the organization.
- Role of IT items in AI integration: IT teams will be the pioneers who drive the integration of generative AI within their companies, and it is essential that the team can show how these applications benefit other groups within the organization. Begin with automating repetitive tasks within your team to exhibit how generative AI can be used to optimize workflows. Familiarize the marketing group, for example, with how tasks such as creating pitches for specific products can be done through ChatGPT. It will obligingly comply if you tell ChatGPT to create three pitches for buying high-performance PCs — or any similar product. And IT leaders are already developing generative AI models targeted specifically at the marketing function.
- Educate the key stakeholders: Aware business leaders about generative AI’s potential benefits and uses through workshops, seminars, and training sessions that help employees upskill and understand how to work with generative AI tools and technologies. Identify and promote real-life use cases within your IT team that help drive the objectives of your organization. Surprise people by beginning with something as simple as using generative engines to craft emails and employee communications.
- Identify applicable use cases: By forming a small task force to become your generative AI experts, locate use cases. Your most enthusiastic people will jump at the opportunity. Have them look for processes or tasks within your current tech stack that can be automated or enhanced through natural language processing, image generation, computer vision, or recommendation systems. Test these concepts using low-risk “fail fast” techniques that leverage small projects and quick feedback. The projects you choose should have a clear scope, defined objectives, and measurable outcomes. Shout about your rooftop successes to gain buy-in and support from key stakeholders.
- Measuring performance and impact: Establish key performance indicators to measure the impact of generative AI solutions on your internal team projects and the overarching business outcomes across departments. Collaborate with other leaders in your company to track metrics such as cost savings, productivity improvements, customer satisfaction, and impact on revenue. Look for quick wins; they’re the best way to bring others on board.
Encourage continuous learning and experimentation across your team. Communicate to others in the organization that AI is the future, and their career success will be enhanced by learning everything they can about applying it.
See Nore: AI Unleashed: A Guide to Responsible Implementation
Growing on Solid Ground
We likely won’t see the real payoffs and implications of using emerging AI technologies in observability and other industries until late 2024 or 2025. Time will tell how much being first to market matters or where customers’ priorities lie.
Are they focused solely on the quality of responses? Speed? Are they comfortable with a fast assistant getting them halfway to a correct answer, knowing they’ll need to maintain a level of suspicion when reviewing answers? Is it better for an AI assistant to do three things well rather than 10 with mediocrity?
GPT-4 remains the gold standard for quality in generative AI. Still, OpenAI has been in the spotlight for turning its attention toward building products around its LLMs, toward profits, and straying from a laser focus on quality. IT leaders must strike the right balance as they embark on their generative AI journeys.
The elephant in the room is how generative AI will impact the workforce. Some jobs will undoubtedly be disrupted in the short term, and history suggests that disruptive technology ultimately creates more jobs than it eliminates.
New technologies invariably grow the labor market pie and create new opportunities where none previously existed. After all, few people could have predicted that 25 years after the launch of Google, there would be 4.2 million job titles on LinkedIn containing the term “SEO.” The entrepreneurs who looked beyond the horizon saw the opportunity for outsized gains.
The same will be true of generative AI.
The first applications of new technologies are usually evident as we seek to improve what we already do. The revolutionary potential comes when we rethink entire processes from the ground up.
Generative AI will soon be as much a standard part of the business landscape as websites and smartphones are today. Technology leaders must look beyond the most apparent use cases to the applications with transformative potential to seize the opportunity. And you must decide one thing right now: “Are you all in or content to dabble around the edges?”
If there’s one thing my 30+ years in tech have taught me, it’s that dabblers aren’t winners. You need to commit yourself and your organization to make generative AI part of the fabric of your business. It needs to be woven into everything your company does.
How have you ensured the smooth integration of GenAI into your business operations? Let us know on Facebook, X, and LinkedIn. We’d love to hear from you!
Image Source: Shutterstock
MORE ON AI IMPLEMENTATION
