Beyond Watson: How enterprises can get AI right with IBM
By Alex Paschal
Senior Systems Engineer
Nordisk Systems, a Converge Company
More than a century ago, Alexander Graham Bell said the immortal words, “Hello, Mr. Watson. Can you hear me?” While he often spoke to his inventor partner, on that particular occasion they were 3,400 miles apart, marking the completion of the first transcontinental phone call – changing the world forever.
The name is now linked in to another tech revolution. This time, it involves IBM’s pioneering work in artificial intelligence (AI), brought into the spotlight by Watson, a supercomputer that utilizes analytical software to create a “question answering” machine. However, like that first phone call, this was just the beginning; AI applications have grown dramatically and enterprises are harnessing the technology.
Here’s how it’s being used by companies, where it’s headed and what you should keep in mind when tackling AI projects.
AI’s driving ambition
With the impact of COVID-19, it’s not surprising Watson has been enlisted to speed up the search for a vaccine. IBM also recently launched Watson Works, a set of products using its AI models and apps to help companies navigate return-to-workplace challenges and slow the spread of the coronavirus.
Still, long before the pandemic, AI had been making major in-roads into the enterprise. The following are just a couple areas and examples of AI business applications now well underway.
● Image recognition: This can enable inspection of components for defects as they roll off the assembly line, reducing costs and increasing customer satisfaction. In healthcare, AI image recognition is helping to process medical images and identify issues like tumors. Self-driving vehicles also rely heavily on image recognition for navigation and safety.
● Data analysis: Using machine learning (ML) with AI allows companies to better predict and anticipate issues. Part replacement demands can be anticipated to avoid downtime. Medical providers can identify which patients will be likely to miss appointments and implement interventions. And, data analysis – in split seconds – is proving crucial for self-driving vehicles.
As for where it’s going, AI essentially uses analytics to compare the past with the present as a kind of “super correlation finder.” There is some fascinating work being done where past-looking AI inferences are used to make predictions, which in turn are re-inferenced to create forward looking AI. Uber’s self-driving AI, which can predict trajectories of pedestrians, vehicles and cyclists, is a good example.
Start small and build
The biggest failure most companies make is jumping into AI and ML without a clear problem to solve. The secret is to start with a single question you need to answer, along with an idea of which dataset you possess that might offer an answer. For instance, a question could be as simple as:
● How do I identify repeat purchase characteristics in customers?
● How can I reduce the defect return rate of products?
AI can provide the answers. And, after you successfully tackle your first project, you’ll find further questions and apps will present themselves. Just keep the scope small to start and get the implementation done, then you can build much more rapidly upon the knowledge. Also, remember that your initial success would likely help you convince decision-makers of AI’s further value.
Go with what you know
Keep in mind that scale is always an issue. It’s not a bad approach to start out with a single person overseeing an AI project, hopefully someone who won’t get bogged down in organizational hurdles. But, eventually, you’ll need to grow your resources and get a team of data scientists working together, which means providing them with the right tools and services.
We always like to go with what we know works best.
That’s why IBM drives all our customer AI initiatives. They’re proven to get results and you know they’ll be there in the future. Equally important, IBM qualifies and works closely with top cloud managed services providers to ensure their technology is used most effectively. For us, that means a particular focus on IBM PowerAI, a complete environment as a service for data science that combines open source deep learning and efficient AI development tools, accelerated by IBM Power Systems.
Read more about IBM Watson
If you’re looking to get AI right, go with what you know works, too. Just send us an email or pick up the phone; we’d welcome the chance to show there’s more to IBM AI than Watson.
Visit Nordisk Systems Advanced Analytics Page for more details.