Technology

Magnetic resonance imaging is an invaluable tool in the medical field, but italso a slow and cumbersome process. It may take fifteen minutes or an hour to complete a scan, during which time the patient, perhaps a child or someone in serious pain, must sit perfectly still. NYU has been working on a way to accelerate this process, and is now collaborating with Facebook with the goal of cutting down MRI durations by 90 percent by applying AI-based imaging tools.

Itimportant at the outset to distinguish this effort from other common uses of AI in the medical imaging field. An X-ray, or indeed an MRI scan, once completed, could be inspected by an object recognition system watching for abnormalities, saving time for doctors and maybe even catching something they might have missed. This project isn&t about analyzing imagery thatalready been created, but rather expediting its creation in the first place.

The reason MRIs take so long is because the machine must create a series of 2D images or slices, many of which must be stacked up to make a 3D image. Sometimes only a handful are needed, but for full fidelity and depth — for something like a scan for a brain tumor — lots of slices are required.

The FastMRI project, begun in 2015 by NYU researchers, investigates the possibility of creating imagery of a similar quality to a traditional scan, but by collecting only a fraction of the data normally needed.

Think of it like scanning an ordinary photo. You could scan the whole thing… but if you only scanned every other line (this is called &undersampling&) and then intelligently filled in the missing pixels, it would take half as long. And machine learning systems are getting quite good at tasks like that. Our own brains do it all the time: you have blind spots with stuff in them right now that you don&t notice because your vision system is filling in the gaps — intelligently.

The data collected at left could be &undersampled& as at right, with the missing data filled in later

If an AI system could be trained to fill in the gaps from MRI scans where only the most critical data is collected, the actual time during which a patient would have to sit in the imaging tube could be reduced considerably. Iteasier on the patient, and one machine could handle far more people than it does doing a full scan every time, making scans cheaper and more easily obtainable.

The NYU School of Medicine researchers began work on this three years ago and published some early results showing that the approach was at least feasible. But like an MRI scan, this kind of work takes time.

&We and other institutions have taken some baby steps in using AI for this type of problem,& explained NYUDan Sodickson, director of the Center of Advanced Imaging Innovation and Research there. &The sense is that already in the first attempts, with relatively simple methods, we can do better than other current acceleration techniques — get better image quality and maybe accelerate further by some percentage, but not by large multiples yet.&

So to give the project a boost, Sodickson and the radiologists at NYU are combining forces with the AI wonks at Facebook and its Artificial Intelligence Research group (FAIR).

NYU School of MedicineDepartment of Radiology chair Michael Recht, MD, Daniel Sodickson, MD, vice chair for research and director of the Center for Advanced Imaging Innovation and Yvonne Lui, MD, director of artificial intelligence, examine an MRI

&We have some great physicists here and even some hot-stuff mathematicians, but Facebook and FAIR have some of the leading AI scientists in the world. So itcomplementary expertise,& Sodickson said.

And while Facebook isn&t planning on starting a medical imaging arm, FAIR has a pretty broad mandate.

&We&re looking for impactful but also scientifically interesting problems,& said FAIRLarry Zitnick. AI-based creation or re-creation of realistic imagery (often called &hallucination&) is a major area of research, but this would be a unique application of it — not to mention one that could help some people.

With a patientMRI data, he explained, the generated imagery &doesn&t need to be just plausible, but it needs to retain the same flaws.& So the computer vision agent that fills in the gaps needs to be able to recognize more than just overall patterns and structure, and to be able to retain and even intelligently extend abnormalities within the image. To not do so would be a massive modification of the original data.

Fortunately it turns out that MRI machines are pretty flexible when it comes to how they produce images. If you would normally take scans from 200 different positions, for instance, itnot hard to tell the machine to do half that, but with a higher density in one area or another. Other imagers like CT and PET scanners aren&t so docile.

Even after a couple years of work the research is still at an early stage. These things can&t be rushed, after all, and with medical data there are ethical considerations and a difficulty in procuring enough data. But the NYU researchers& ground work has paid off with initial results and a powerful data set.

Zitnick noted that because AI agents require lots of data to train up to effective levels, ita major change going from a set of, say, 500 MRI scans to a set of 10,000. With the former data set you might be able to do a proof of concept, but with the latter you can make something accurate enough to actually use.

The partnership announced today is between NYU and Facebook, but both hope that others will join up.

&We&re working on this out in the open. We&re going to be open-sourcing it all,& said Zitnick. One might expect no less of academic research, but of course a great deal of AI work in particular goes on behind closed doors these days.

So the first steps as a joint venture will be to define the problem, document the data set and release it, create baselines and metrics by which to measure their success, and so on. Meanwhile, the two organizations will be meeting and swapping data regularly and running results past actual clinicians.

&We don&t know how to solve this problem,& Zitnick said. &We don&t know if we&ll succeed or not. But thatkind of the fun of it.&

What Robinhood did to democratize buying individual stocks, Titan wants to do for investing in a managed portfolio. Instead of being restricted to rich accredited investors willing to pour $5,000 or even $500,000 into a traditional hedge fund that charges 2 percent fees and 20 percent of profits, Titan lets anyone invest as little as $1,000 for just a 1 percent fee on assets while keeping all the profits. Titan picks the top 20 stocks based on data mined from the most prestigious hedge funds, then invests your money directly in those with personalized shorts based on your risk profile.

Titan has more $10 million under management after quietly spinning up five months ago, and this week the startup graduates from Y Combinator. Now Titan is ready to give upscale millennialsa more sophisticated way to play the markets.

This startup is hot. It refused to disclose its funding, likely in hopes of not tipping off competitors and incumbents to the opportunity itchasing. But itthe buzz of YC, with several partners already investing their own money through Titan. When you consider Stanford-educated free stock-trading app Robinhoodstunning $5.6 billion valuation thanks to its disruption of E*Trade, iteasy to imagine why investors are eager to back Titanattack on other financial vehicles.

&We&re all 28 to 30 years old,& says co-founder Clayton Gardner about his team. &We want to actively invest and participate in the market but most of us who don&t have experience have no idea what we&re doing.& Most younger investors end up turning to family, friends or Reddit for unreliable advice. But Titan lets them instantly buy the most reputable stocks without having to stay glued to market tickers, while using an app to cut out the costs of pricey brokers and Wall Street offices.

Titan co-founders (from left): Max Bernardy, Joe Percoco,Clayton Gardner

&We all came from the world of having worked at hedge funds and private equity firms like Goldman Sachs. We spent five years doing that and ultimately were very frustrated that the experiences and products we were building for wealthy people were completely inaccessible to people who weren&t rich or didn&t have a fancy suit,& Gardner recalls. &Instead of charging high fees, we can use software to bring the products directly to consumers.&

How Titan works

Titan wants to build BlackRock for a new generation, but its origin is much more traditional. Gardner and his co-founder Joe Percoco met on their first day of business school at UPennWharton (of course). Meanwhile, Titanthird co-founder, Max Bernardy, was studying computer science at Stanford before earning a patent in hedge fund software and doing engineering at a few startups. The unfortunate fact is the world of finance is dominated by alumni from these schools. Titan will enjoy the classic privilege of industry connections as it tries to carve out a client base for a fresh product.

&We were frustrated that millennials only have two options for investing: buying and selling stocks themselves or investing in a market-weighted index,& says Gardner. &We&re building the third.&

Titanfirst product isn&t technically a hedge fund, but itbuilt like one. It piggybacks off the big hedgies that have to report their holdings. Titan uses its software to determine which are the top 20 stocks across these funds based on turnover, concentration and more.All users download the Titan iOSor Android app, fund their account and are automatically invested into fractional shares of the same 20 stocks.

Titan earns a 1 percent annual fee on what you invest. There is a minimum $1,000 investment, so some younger adults may be below the bar.&We&re targeting a more premium millennial for start. A lot of our early users are in the tech field and are already investing,& says Gardner.

For downside protection, Titan collects information about its users to assess their risk tolerance and hedge their investment by shorting the market index 0 to 20 percent so they&ll earn some if everything crashes. Rather than Titan controlling the assets itself, an industry favorite custodian called Apex keeps them secure. The app uses 256-bit encryption and SSL for data transfers, and funds are insured up to $500,000.

How have its bets and traction been doing &We&ve been pleasantly surprised so far,& Gardner beams, noting Titanthousands of clients. It claims itup 10 percent year-to-date and up 33 percent in one year compared to the S-P 5002 percent year-to-date and 22 percent in one year. Since users can pull out their funds in three to four business days, Titan is incentivized to properly manage the portfolio or clients will bail.

But beyond the demographic and business model, itthe educational elements that set Titan apart. Users don&t have to hunt online for investment research. Titan compiles it into deep dives into top stocks like Amazon or Comcast, laying out investment theses for why you should want your money in &the everything store& or &a toll road for the Internet.& Through in-app videos, push notifications and reports, Titan tries to make its users smarter, not just richer.

With time and funding, &Eventually we hope to launch other financial products, including crypto, bonds, international equities, etc.,& Percoco tells me. That could put Titan on a collision course with Wealthfront, Coinbase and the recently crypto-equipped Robinhood, as well as direct competitors like asset managers BlackRock and JP Morgan.

&If we fast-forward 10 to 20 years in the future, millennials will have inherited $10 trillion, and at this rate they&re not equipped to handle that money,& says Gardner. &Financial management isn&t something taught in school.&

Worryingly, when I ask what they see as the top threats to Titan, the co-founders exhibited some Ivy League hubris, with Gardner telling me, &Nothing that jumps out…& Back in reality, building software that reliably prints money is no easy feat. A security failure or big drop could crater the appbrand. And if its education materials are too frothy, they could instill blind confidence in younger investors without the cash to sustain sizable losses. Competitors like Robinhood could try to swoop in an offer managed portfolios.

As a teenager in Nigeria, I tried to build an artificial intelligence system. I was inspired by the same dream that motivated the pioneers in the field: That we could create an intelligence of pure logic and objectivity that would free humanity from human error and human foibles.

I was working with weak computer systems and intermittent electricity, and needless to say my AI project failed.Eighteen years later— as an engineer researching artificial intelligence, privacy and machine-learning algorithms — I&m seeing that so far, the premise that AI can free us from subjectivity or bias is also disappointing. We are creatingintelligence in our own image. And thatnot a compliment.

Researchers have known forawhilethat purportedly neutral algorithms can mirror or even accentuate racial, gender and other biases lurking in the data they are fed. Internet searches on names that are more often identified as belonging to black people were found to prompt search engines to generate ads for bail bondsmen. Algorithms used for job-searching were more likely to suggest higher-paying jobs to male searchers than female. Algorithms used in criminal justice also displayed bias.

Five years later, expunging algorithmic bias is turning out to be a tough problem. It takes careful work to comb through millions of sub-decisions to figure out why the algorithm reached the conclusion it did. And even when that is possible, it is not always clear which sub-decisions are the culprits.

Yet applications of these powerful technologies are advancing faster than the flaws can be addressed.

Recent researchunderscores this machine bias, showing that commercial facial-recognition systems excel at identifying light-skinned males, with an error rate of less than 1 percent. But if you&re a dark-skinned female, the chance you&ll be misidentified rises to almost 35 percent.

AI systems are often only as intelligent — and as fair — as the data used to train them. They use the patterns in the data they have been fed and apply them consistently to make future decisions. Consider an AI tasked with sorting the best nurses for a hospital to hire. If the AI has been fed historical data — profiles of excellent nurses who have mostly been female — it will tend to judge female candidates to be better fits. Algorithms need to be carefully designed to account for historical biases.

Occasionally, AI systems get food poisoning. The most famous case wasWatson, the AI that first defeated humans in 2011 on the television game show Jeopardy. Watsonmasters at IBM needed to teach it language, including American slang, so they fed it the contents of the online Urban Dictionary. But after ingesting that colorful linguistic meal, Watson developed a swearing habit. It began to punctuate its responses with four-letter words.

We have to be careful what we feed our algorithms. Belatedly, companies now understand that they can&t train facial-recognition technology by mainly using photos of white men. But better training data alone won&t solve the underlying problem of making algorithms achieve fairness.

Algorithms can already tell you what you might want to read, who you might want to date and where you might find work. When they are able to advise on who gets hired, who receives a loan or the length of a prison sentence, AI will have to be made more transparent — and more accountable and respectful of societyvalues and norms.

Accountability begins with human oversight when AI is making sensitive decisions. In an unusual move,Microsoft president Brad Smithrecently called for the U.S. government to consider requiring human oversight of facial-recognition technologies.

The next step is to disclose when humans are subject to decisions made by AI. Top-down government regulation may not be a feasible or desirable fix for algorithmic bias. But processes can be created that would allow people to appeal machine-made decisions — by appealing to humans. The EUnewGeneral Data Protection Regulationestablishes the right for individuals to know and challenge automated decisions.

Today people who have been misidentified — whether in an airport or an employment data base — have no recourse. They might have been knowingly photographed for a driverlicense, or covertly filmed by a surveillance camera (which has a higher error rate). They cannot know where their image is stored, whether it has been sold or who can access it. They have no way of knowing whether they have been harmed by erroneous data or unfair decisions.

Minorities are already disadvantaged by such immature technologies, and the burden they bear for the improved security of society at large is both inequitable and uncompensated. Engineers alone will not be able to address this. An AI system is like a very smart child just beginning to understand the complexities of discrimination.

To realize the dream I had as a teenager, of an AI that can free humans from bias instead of reinforcing bias, will require a range of experts and regulators to think more deeply not only about what AI can do, but what it should do — and then teach it how.