To fans of science fiction, artificial intelligence may remind them of robots like C-3PO, the loquacious but harmless golden droid in Star Wars, or Skynet in the Terminator movies, a calculating sentient computer that subjugated mankind. But AI is more than just a machine with human-level intelligence scientists hope they could one day create. It is a set of algorithms and technologies that is already powering many tasks in everyday life.
When Google Photos groups images of people using facial recognition, it deploys the deep learning techniques of AI. Chatbots that converse with you in Yahoo, Facebook and other sites use AI. Alibaba harnesses deep learning to find a handbag matching the one in the photo you uploaded to its shopping site. Digital assistants like Siri, Alexa, Cortana and Google Assistant use AI to provide information or execute tasks.
The advancements continue. Computers are now able to generate an image that 40% of people could mistake for a photographed face, according to Lukasz Kaiser, senior research scientist at Google Brain, who spoke at the recent AI Frontiers conference in Silicon Valley. Previously, computers could only fool about a tenth of the people. "We're very close to generating photos that look like real faces." AI can also take search queries and reference web pages to create an article, such as a person's biography, based on everything about him or her found on the internet, he said.
More companies are jumping into AI development. Frank Chen, a partner at top venture capital firm Andreessen Horowitz, said many of the roughly 1,500 startups his firm sees annually began turning to AI two years ago. "Now, 60% to 70% of the companies we see self-identify as an AI startup," he said at the conference. Chen predicted that AI will become as pervasive in business as databases did because of its usefulness. Databases are "so generally useful, they got assigned to all the applications. AI is exactly the same," he added. "AI is going to get into everything."
"[Databases are] so generally useful, they got assigned to all the applications. AI is exactly the same."
Chen added that investors will soon stop looking specifically for AI startups, and instead assume that all startups are using AI in some form. It's the same trajectory that mobile technology and cloud computing followed. In 2009, being a cloud native was "new and different," and startups also began distancing themselves from desktops to identify as mobile-first. "Now, if we see a startup that's not cloud native and not mobile-first, we ask, 'What's going on with you?' AI is going through the same exact thing," he said.
Startups deploying AI can do everything from using drones for medical deliveries to helping lawyers prepare for court. Chen said his firm has invested in Zipline, an AI startup that uses drones to deliver blood to remote places such as western Rwanda. The service is critical to locations that are hard to access by land. "By the time a truck can get there, it may be too late," he said. Medical personnel in the field use an app to order blood by type. Half an hour later, a drone drops it out of the sky. The drone has a 75-kilometer range and accuracy of five feet. Chen said Zipline is making up to 500 delivers a day and is looking to expand.
Everlaw helps lawyers prepare for trials. The first step in any trial is the gathering of evidence. The startup uses AI to do things like read documents to find ones helpful to the lawyer's case and identify those that need to be sent to the opposition to avoid a mistrial. "This categorizes documents automatically so you don't miss documents that are important," Chen said.
Naturali is a China-based firm that uses AI and speech recognition to speed up the access and usage of apps. For example, to order an Uber ride, you have to unlock your phone, click on the Uber app, type in the destination, choose the type of ride and then book it. "Wouldn't it be nice if you can just tell your phone - 'Uber ride Crowne Plaza San Francisco' - and then the Uber app just gives you the car?" said Dekang Lin, Naturali's co-founder and chief technology officer. "We upgrade apps with a speech interface ... to translate speech into clicks and swipes."
Startups deploying AI can do everything from using drones for medical deliveries to helping lawyers prepare for court.
aims to make meetings more productive. It created Eva, a digital personal assistant for the workplace. Omar Tawakol, founder and CEO of the startup, said the most used forms of business collaboration are instant messaging and email. But during meetings, employees are disconnected from this workflow and may forget to do follow-ups. Moreover, "the vast majority of the conversations are lost in the ether," he said. Eva will take notes, process the meeting and send a summary of it to participants. It also could be instructed to do action items, such as send a copy of a presentation to everyone in the meeting.
AI is a startup that developed drones to fly through combat or crisis situations. For example, the drones can map a building in real time to construct 3D maps and identify the people inside. "Today, when we're in hostile territory, we send young people with guns inside buildings," Chen said.
IoT helps companies solve business problems by leveraging four big tech trends: big data, cloud computing, AI and machine learning as well as the Internet of Things. For example, it helped one Italian utility detect fraud using smart meters. About 3% of electricity in that country is stolen, said Zico Kolter, chief data scientist. Some people even put car cables around the electricity meter to siphon it off. "We predict the probably of fraud and probability of recovering energy," he said. In another case, the startup monitored oil and gas wells to see which are more likely to fail and cause environmental issues.
"Wouldn't it be nice if you can just tell your phone - 'Uber ride Crowne Plaza San Francisco' - and then the Uber app just gives you the car?" - Dekang Lin
is a life extension startup that uses machine learning to help people live longer. "They look for small molecules in your blood stream that are predictive of mortality," said Chen, whose firm has invested in the company. "What they've done through machine learning is identify... molecules to focus on for drug discovery that might help us extend life."
is a drone analytics startup serving industries such as mining. For example, mine owners need to follow many safety regulations and one rule is to use rocks as road markers to guide vehicles. The rule is that these rocks need to be twice as high as the tallest wheel of the vehicle. Mine owners usually send people out to measure the height of the rocks. Airware simplifies the process by using drones to analyze imagery and identify rocks that are too short. It can also analyze a road's steepness and recommend that drivers take another route to save on fuel.