It started with a hunch.
After spending years at the University of Michigan studying and designing algorithms — the mathematical models that operate any computerized process — Mehrzad Samadi began exploring the fields he thought might be best suited for a unique artificial intelligence (AI) startup.
Soon after, he and fellow U-M grad Ankit Sethia settled on genomics, an expansive medical discipline that uses genetic information about an individual to design better clinical care options and aid in diagnostic decision-making.
Together the duo would go on to license the AI technology, which is capable of receiving and reacting to vast amounts of data in real time, and launched Ann Arbor-based Parabricks in 2015. The enterprise uses reactionary software to rapidly accelerate the virtual process of analyzing genetic information extracted from patients, allowing for faster diagnoses and treatments of an assortment of ailments and diseases.
“Our goal is to go from data to insight in the shortest time period,” says Samadi, CEO of Parabricks. “We enable our customers to use machine learning (and AI) on large amounts of data to come up with insights that were hidden in data with traditional applications.”
Hospitals and diagnostic labs are often limited by a combination of high operational costs and the excessive time lapses involved with processing an entire genome. That phenomenon has led to the routine use of DNA that’s, at best, limited — and, at worst, woefully insufficient for diagnosing disease.
Parabricks counters those drawbacks by utilizing raw genome data received from a virtual cloud to generate full genetic reports in a timeframe few working in the genomics field would’ve imagined possible. “We finish computations that take two days on a normal server in less than one hour,” Samadi says.
“You’re likely to see highly usable AI in the next few years. It’s likely to be pretty hidden. It’ll be doctors using (AI) to read X-rays, using email (networks that) auto-matically filter spam, and other applications.”
— Ian Sefferman, Assembler Labs
In the not too distant future, Samadi hopes to enhance Parabricks’ software capacity by offering whole genome sequencing options and follow-up data analysis for a broad spectrum of patients, particularly in the Southeast Asia genomics market. “The market is growing fast and they’re looking for innovative solutions that can help them move forward,” he says. “That’s why Southeast Asia is one of the main targets for our technology.”
At the same time, Loven Systems, a cognitive analytics solutions company in Northville Township, is continuing to attract interest and gain clients representing multiple industries following its 2017 release of Diwo — short for “data in, wisdom out” — a cognitive decision-making software platform that utilizes AI.
Launched at the Strata Data Conference in New York last September, Diwo proactively guides business decision-making by virtually reading live data streams in a predictive manner to identify revenue opportunities, patterns, and trends before they occur.
In essence, Diwo’s sixth-sense AI capacity provides logistical and financial incentives from the first time clients put the software platform to use.
“Diwo augments the user’s unique decision-making process by providing actionable, interactive insights in their own business context,” says Dana Turcas, a marketing principal at Loven Systems. “This cognitive decision-making framework also sets itself apart with its business-first approach, as we’re keenly aware of adoption issues common with most transformational initiatives. By leveraging your existing technologies and providing an end-to-end solution, Diwo provides value on day one.”
She says the company’s customer base comes from the retail, health care, financial services, insurance, e-commerce, automotive, and gaming sectors.
In turn, client feedback regarding Diwo was so positive post-launch that Loven Systems developed a scaled-back version of its flagship product in June. DiwoPronto, whether used in a stand-alone manner or integrated with Diwo, was designed for clients that may require real-time analytic reads and decision-making assistance for less complex business operations.
To casual observers, Parabricks and Loven Systems might represent two Michigan-based companies that have successfully tapped into what’s actually possible with AI, a technology some are convinced will one day routinely perform feats tantamount to miracles.
Their products are reactionary in real time and, most importantly, have the capacity to continuously learn over time.
What they’ve been able to achieve is certainly worthy of accolades. Unfortunately, though, as circumstances have it, any high praise in the AI world is often fleeting.
Even when taking into account the innovation that each company possesses, the commercial AI product deployments from Parabricks and Loven Systems only allow each information technology company to keep pace in the global race that’s only limited by the imagination.
More simply put, for those racing to make inroads in the evolving international AI arena — and China is expending billions, in a clear play for dominance — success is an intricately designed algorithm away from becoming reality.
Add to that the fact that estimates have been made claiming that AI could potentially generate anywhere between $3.5 trillion and $5.8 trillion in new revenue across multiple
industries globally, and it’s easy to see why the field has no shortage of innovators attempting to seize opportunities, assume risks, and take chances.
The wide-open opportunity to completely set themselves apart in AI development was likely the reason why Google was willing to leave attendees either awestruck or frightened at its annual I/O developer festival this past May, when they unveiled their latest version of a virtual assistant, Duplex, which can schedule appointments over the phone using perfectly replicated human male and female voices.
Overseas, AI developers at the Israeli technology firm Cortica are digitally mapping and modeling the cortex regions found in rat brains, in hopes that they’ll unlock a level of reasoning capacity that could be used to design everything from surveillance systems to autonomous vehicle software interfaces.
In metro Detroit, the opportunity to become the world’s foremost leader in autonomous vehicle development was a driving factor behind Ford Motor Co.’ s $1 billion investment last year in Argo AI, a Pittsburgh-based technology company founded by former leaders of Google and Uber.
“Our partnership with Argo AI is going very well… They’ve built a very talented team of more than 325 people so far, and are testing self-driving vehicles in and around (Dearborn, Miami, and Pittsburgh).” — Alan Hall, Communications Manager for Ford’s Autonomous and
Electric Vehicles Research and Advanced Engineering Office
Spread out over five years, the investment is blending the Dearborn automaker’s autonomous vehicle design expertise with Argo AI’s robotic and software engineering capabilities. Uniquely structured, the arrangement — which centers on developing a software platform capable of being used on Ford’s first commercial autonomous vehicle model, slated for 2021 — provides Argo AI with an equity stake and licensing rights.
Alan Hall, communications manager for Ford’s autonomous and electric vehicles research and advanced engineering office, says the partnership with Argo AI is meeting expectations. “Our partnership with Argo AI is going very well,” Hall says. “They’ve built a very talented team of more than 325 people so far, and are testing self-driving vehicles in and around (Dearborn, Miami, and Pittsburgh).”
With respect to Ford’s recent acquisition of the long-closed Michigan Central Station, Hall says plans call for creating an AI footprint near and around the site, once the restoration allows (the building is projected to open in 2022).
“In terms of the Michigan Central Station, we expect that our mobility team will make up a large majority of employees who will eventually be based there,” Hall says. “(We’re) working to infuse AI technology in many facets of our business, so as we get closer to the move-in day, we’ll spell out the specifics of what the teams will be responsible for in Detroit.”
Its crosstown rival, General Motors Co. in Detroit, is striving to beat Ford to the punch in the commercial autonomous vehicle market by utilizing a recent $2.5 billion investment from SoftBank Group Corp. and $1.1 billion of its own funds to further bankroll Cruise, the automaker’s autonomous vehicle unit that’s planning to make a self-driving car with no steering wheels or pedals available for purchase next year.
Apart from automotive mobility projects, multiple industries from banking, health care, marketing, real estate, and security now incorporate some form of AI to varying degrees. “AI has applications in virtually anything you can think of,” says Ian Sefferman, a metro Detroit information technology industry veteran experienced in application development.
Whether it’s Netflix recommending a movie or a show, an online retailer like Amazon suggesting products or services to purchase, or social media channels making friend and follower suggestions, there’s an algorithm operating somewhere in a virtual background, making informed decisions based on incoming data — personal or otherwise.
Sefferman, who recently cofounded Assembler Labs, a Detroit startup dedicated to fostering venture-scale business opportunities in the state, sees the presence and overall use of AI increasing exponentially in both the immediate and long-term future. “You’re likely to see highly usable AI in the next few years,” he says. “It’s likely to be pretty hidden. It’ll be doctors using (AI) to read X-rays, using email (networks that) automatically filter spam, and other applications.”
Analysis regarding how AI may influence businesses in the state going forward is the subject of much debate. According to Tom Kelly, executive director of Automation Alley in Troy, members of the technology and manufacturing consortium aren’t shying away from AI technology.
“A lot of companies are seeking knowledge,” Kelly says. “(They’re primarily) asking about sensors, 3-D printing, and robotics.”
Still, Kelly acknowledges the current high cost of utilizing AI is a mitigating factor that, so far, doesn’t have a solution. “Smaller companies are much less able to do those kinds of (AI) projects,” Kelly says.
“The market is growing fast and they’re looking for innovative solutions that can help them move forward.” — Mehrzad Samadi, CEO of Parabricks
In a big picture perspective, Kelly takes a “counterintuitive approach” to those predicting that increases in AI automation will lead to job losses and slow growth. “Labor is expected to be flat in the next 40 years,” he says. “If you want to grow GDP, you must automate.”
Well beyond Michigan, a series of other high-tech leaders are rapidly integrating AI into numerous projects, from Microsoft founder Bill Gates to Elon Musk of SpaceX and Tesla fame to Jeff Bezos of Amazon.
Even for all of its promise, some AI users like Musk caution about what has come to be known as the “master algorithm.” Theoretically, at some point in the future an algorithm could successfully be developed that has the capability to learn from any type of data in the world. From there, it can virtually perform any task, regardless of its complexity, without any kind of human intervention. It could also bypass human input altogether.
Worst yet, AI could make the world go completely awry if the technology eventually advances to the point that algorithms begin designing their own codes. “An AI algorithm (that becomes) smart enough to design its own algorithms would be problematic,” Sefferman says.
Provided that it’s developed with enough accurate data at the onset, AI may eventually be able to seamlessly diagnose all types and stages of cancer; however, any predictions made won’t come with a semblance of the empathic bedside manner people around the world have become accustomed to.
Such potential, combined with a clear lack of empathy, distinguishes AI in one very distinct way. For all of its future promise, AI is a technology chock-full of irony.
More specifically, AI capability, no matter how uniquely targeted or highly sophisticated in its intended scope, entirely hinges on algorithms specially designed to go only as far as any developer, or team of developers, so pleases.
With that, one all-important question in Michigan is how will AI transform certain aspects of an industrial blue-collar base for a more automated future? And ultimately, how much will AI irreversibly change the business landscape in Michigan? Only time will tell.