Welcome to AI Decoded, Fast Company’s weekly newsletter that breaks down the most important news in the world of AI. You can sign up to receive this newsletter every week here.
The Pentagon needs Silicon Valley’s AI
The Pentagon is finally getting better at procuring technology from Silicon Valley. Private-sector tech companies, especially small ones, have often had difficulty selling their wares to the Pentagon thanks to bureaucratic red tape and excruciatingly slow sales cycles. Wars in Ukraine and Syria have made it increasingly clear that the Pentagon needs the tech sector’s innovations to counter 21st century battlefield threats. And that new approach is really bearing fruit when it comes to artificial intelligence.
One Pentagon source tells me that while the Defense Department (DoD) has not completely solved the problem of acquiring goods from tech startups, it has still come a long way. It has established a whole community of “innovation centers,” the biggest of which is the Defense Innovation Unit (DIU), to help tech startups get their products and practices ready to serve the Pentagon as a client. Increasingly, these are AI companies or companies that use AI in their products. Some themes have emerged in the ways the DoD wants to apply AI.
It’s no secret that the wars in Syria and Ukraine have taught the Pentagon a lot about the kinds of threats it could face if pulled into a new conflict.The DoD and the DIU are particularly concerned about fighting battles in “denied environments” where normal communication systems have been shut down. In such a scenario, an operator’s control over a drone might be disrupted, making an autonomous drone—one with a large enough AI model onboard to carry out a mission even if its communication channels are lost—all the more valuable. Ukraine and NATO recently held a hackathon in Poland to address this very problem.
The Defense Department is focused on building up the infrastructure needed to host and support its own AI models. Beginning in 2022 the DIU worked with private-sector players to connect the DoD’s on-premise supercomputers with a wide range of computing resources offered by commercial cloud computing providers. This creates more computing power for DoD researchers, access to more types of hardware, and more flexibility in accessing computing power.
The Pentagon uses AI for a wide variety of things, including tasks like equipment maintenance that are far from the battle. The most pressing needs are technologies that could save soldiers’ lives. So a lot of money is flowing into startups developing new methods of detecting and destroying autonomous weapons.
Arianna Huffington on her new AI health venture with Sam Altman
Arianna Huffington announced this week that she’s teamed up with OpenAI CEO Sam Altman to form a new company called Thrive AI Health, which will build an AI-powered “hyper-personalized” AI health coach to help people manage chronic conditions and live healthier lives. The “coach” will first come in the form of a platform that enterprises offer to employees, and later as a consumer app. The new service will bring to bear existing behavioral health tools developed by Huffington’s company, Thrive Global, and combine it with the power of an OpenAI model, which will generate the personalized coaching content. (The system will also be trained on relevant health research from places like Stanford.) The initiative will be partly bankrolled by the OpenAI Startup Fund, which provides funding for young companies that are building apps and services atop OpenAI models. DeCarlos Love, who was the health and wearables product leader at Google, will serve as CEO of the new company.
Thrive Global already had a health-tech platform used by enterprises to help employees stay healthy (and less likely to utilize expansive healthcare services), but the advent of generative AI opened the door for a service that could learn a lot more about users’ habits and tastes, and give them proactive strategies for overcoming temptations and making better choices, Huffington tells me.
“One of the things that we’re really excited about is the health-equity part of this, how we can democratize and scale what people with resources already know and many practice,” Huffington says, referring to the “wealthy and well” population who often use Fitbits or Apple Watches to monitor everything from sleep to food to movement. Meanwhile, the people who really need the help—less-affluent people with diseases, such as obesity, or disabilities—often can’t afford fancy gadgets or personal health coaches. “We work with a lot of people who don’t have that kind of information or that kind of support to make changes in their lives,” she says.
Huffington explains the new solution, which will be an enterprise service and a mobile app, will be hardware agnostic, and will ingest whatever biometrics information the user authorizes it to use. She adds that the AI coach, while very well-informed, will not give out medical advice (apps that usually do this require FDA approval).
Could deepfakes be a no-show in the 2024 election?
AI may not be the existential threat to the integrity of 2024’s elections that people have been fearing. Yes, we’ve seen a few deepfake images and videos in elections abroad, and we’ve seen one (poor) deepfaked audio message from Joe Biden. But we’ve not yet seen widespread use of deepfakes in order to smear competing candidates or causes.
“I don’t know if deepfakes are going to be as big a deal as people have thought, simply because we’re quite capable of lying without it,” says political consultant Colin Delany. Delany says bad actors (known as “ratf*ckers” in campaign circles) don’t need anything so fancy as AI to mislead people. “Selectively edited videos do not require AI,” he says. Bad actors also often repurpose legit-but-old images or footage as evidence of current claims.
Delany suggests that people may be on guard against deepfakes in part because the media has covered the tech so relentlessly (Fast Company included). Given the low levels of consumer trust of consumer AI products like ChatGPT, and the well-known potential for using AI to spread information (“mis-” and otherwise), people’s willingness to take any digital content at face value may have dropped.
More AI coverage from Fast Company:
- AI startup funding doubled in Q2, Crunchbase data shows
- AI vulnerabilities to space pirates: Cyberattacks in space can threaten life on Earth
- The worst branding blunders of the AI era—so far
- I’m a venture capitalist, and these are the 3 AI myths that everyone needs to stop believing
Want exclusive reporting and trend analysis on technology, business innovation, future of work, and design? Sign up for Fast Company Premium.