How do you use AI to disrupt your business? What do you need to do in order to find value, justify the expense, create an impact on your business?
These are some of the questions that were asked and answered at VB Summit 2018. Many executives from global companies like Microsoft, Google, Deloitte, and Uber, came to share how they are transforming their companies internally and helping customers and partners achieve more by implementing artificial intelligence.
Day 1 Key Takeaways
The first day of VB Summit 2018 was focused on the current state of AI in businesses. Companies including Berg Health and Pandora spoke about how AI is changing their businesses. In fact, there are a couple of great stories worth sharing.
#1 Why use AI at all? What value does it bring?
Many people think AI is an all-in-one solution. In reality, AI is not a packaged solution but a tool to empower your business. Slava Akmaev, SVP/Chief Analytics Officer at BERG Health, shared a great analogy that explained why his company embraces AI in their business.
“Imagine that you buy a new iPhone but find it broken when you got home. You bring it back to the Apple store, but they tell you that the only solution is to throw the broken phone away and get a new one. It’s about $1,000, but you are okay with the expense. Now let’s change the iPhone with a Tesla. You buy a new Tesla, but it doesn’t start. The only solution is to dump that and get a new one. You may have been okay with getting two iPhones but how about buying two Teslas? Finding a cancer treatment is even more expensive, but this is what happens. You work so hard to find a cure, but if it doesn’t work, all that hard work gets thrown away, and you have to start all over again. That’s why speed is critical. We need to be able to shorten the cycle of trial and error so that we minimize the expense and increase the chance of finding effective treatments, and we are achieving this by using AI to accelerate the process.”
BERG’s case may sound unique, but it really isn’t. Every business has to invest for the future, and sometimes historical data is insufficient to make these decisions with complete accuracy. However, companies that wait for more accurate predictions risk losing competitive advantage. They could have been first-movers, but because they waited, became followers and lost market share. What more can you do if you can make more confident decisions faster?
#2 Lack of talent in the market
“A 3rd grader was walking back home and saw three workers laying bricks. She asks the first worker what he’s doing, and he says ‘What do you think I’m doing? Laying bricks!’ She moves on and asks the second worker, and he responds ‘I am laying bricks to make money.’ She nods and asks the last worker, and he says ‘I’m building a cathedral.'”
The point of the story was that you need to hire someone who has a sense of purpose, not for the sake of the job or for the money. Having purpose is extremely important, Sid added, because you don’t want a data scientist to just build a model. You want someone who can elaborate on why they built the model, how they built it, and what values it will bring to the company. For BERG, the value was speed. What questions do you ask in interviews and what values do you look for?
Day 2 Key Takeaways
The second day was more about applying AI in practice. For example, Dr. Andrew Siemion, Director at Berkeley SETI Research Center, explained how they were able to identify 72 possible extraterrestrial pulses by using AI in a month which they couldn’t possibly have otherwise. In the “Titans in AI” panel, the attendees shared stories about how every business will need to implement AI soon, but the primary objective should always focus on the business impact. A great example from Venky Veeraraghavan, Group PM for Azure Cloud, Microsoft, was that PowerPoint definitely uses AI but PowerPoint is still a PowerPoint with improvements, not something else.
If adopting AI is inevitable, should you build or buy AI?
The Build or Buy Conundrum breakout session brought experts – such as Pedro Alves, our founder and CEO, Michael Butler, Director of Global eCommerce at VMware, Sarah Aerni, Director of Data Science & Engineering at Salesforce, and Jairam Ranganathan, Director of Product at Uber – together to discuss the ups and downs of delivering AI within their companies.
The panelists all had multiple years of experience building their own data science teams and models, – as well as buying AI solutions from vendors. They spoke about the different criteria to consider when making the decision to build or buy AI. For example, Jai explained that he chose to build in-house because there wasn’t a vendor that could scale with Uber to handle big data. Another consideration discussed was “how many partners are enough?”. Today, no single vendor has all the AI components an enterprise requires. The panel warned about how complicated it can be to even start evaluating multiple vendors to work with in concert.
More detailed story on this breakout session will be updated soon. Stay tuned! Click here for the breakout session summary)
In short, the attendees were able to reach four big lessons.
- Buy whenever possible
- Each team should have access to AI, but learnings need to be shared across all teams
- Domain experts need to drive – define the problems and (as much as possible) use the AI tools
- Make sure you have the people and structures in place to use the results from AI models
The two days we spent at VB Summit 2018 were a very exciting experience for us, especially with the Series A funding news exclusively featured on VentureBeat. We met new people and learned a lot about how we can further improve our great product to fit different needs. Special thanks to Matt Marshall, Gina Joseph, and Dilan Yuksel for guiding us throughout the event! Now let’s watch Pedro’s toast after the successful two days.