I’m back from a healthful week off picking blueberries in a forest. So this story we published last week in regards to the messy ethics of AI in warfare is just the antidote, bringing my blood pressure right back up again. 

Arthur Holland Michel does an important job taking a look at the complicated and nuanced ethical questions around warfare and the military’s increasing use of artificial-intelligence tools. There are myriad ways AI could fail catastrophically or be abused in conflict situations, and there don’t appear to be any real rules constraining it yet. Holland Michel’s story illustrates how little there’s to carry people accountable when things go flawed.  

Last yr I wrote about how the war in Ukraine kick-started a brand new boom in business for defense AI startups. The newest hype cycle has only added to that, as corporations—and now the military too—race to embed generative AI in services and products. 

Earlier this month, the US Department of Defense announced it’s organising a Generative AI Task Force, geared toward “analyzing and integrating” AI tools resembling large language models across the department. 

The department sees tons of potential to “improve intelligence, operational planning, and administrative and business processes.” 

But Holland Michel’s story highlights why the primary two use cases could be a nasty idea. Generative AI tools, resembling language models, are glitchy and unpredictable, and so they make things up. Additionally they have massive security vulnerabilities, privacy problems, and deeply ingrained biases.  

Applying these technologies in high-stakes settings could lead on to deadly accidents where it’s unclear who or what ought to be held responsible, and even why the issue occurred. Everyone agrees that humans should make the ultimate call, but that’s made harder by technology that acts unpredictably, especially in fast-moving conflict situations. 

Some worry that the people lowest on the hierarchy can pay the best price when things go flawed: “Within the event of an accident—no matter whether the human was flawed, the pc was flawed, or they were flawed together—the one that made the ‘decision’ will absorb the blame and protect everyone else along the chain of command from the total impact of accountability,” Holland Michel writes. 

The one ones who seem more likely to face no consequences when AI fails in war are the businesses supplying the technology.

It helps corporations when the foundations the US has set to control AI in warfare are mere recommendations, not laws. That makes it really hard to carry anyone accountable. Even the AI Act, the EU’s sweeping upcoming regulation for high-risk AI systems, exempts military uses, which arguably are the highest-risk applications of all of them. 

While everyone seems to be in search of exciting latest uses for generative AI, I personally can’t wait for it to develop into boring. 

Amid early signs that folks are beginning to lose interest within the technology, corporations might find that these kinds of tools are higher fitted to mundane, low-risk applications than solving humanity’s biggest problems.

Applying AI in, for instance, productivity software resembling Excel, email, or word processing may not be the sexiest idea, but in comparison with warfare it’s a comparatively low-stakes application, and easy enough to have the potential to truly work as advertised. It could help us do the tedious bits of our jobs faster and higher.

Boring AI is unlikely to interrupt as easily and, most vital, won’t kill anyone. Hopefully, soon we’ll forget we’re interacting with AI in any respect. (It wasn’t that way back when machine translation was an exciting latest thing in AI. Now most individuals don’t even take into consideration its role in powering Google Translate.) 

That’s why I’m more confident that organizations just like the DoD will find success applying generative AI in administrative and business processes. 

Boring AI shouldn’t be morally complex. It’s not magic. Nevertheless it works. 

Deeper Learning

AI isn’t great at decoding human emotions. So why are regulators targeting the tech?

Amid all of the chatter about ChatGPT, artificial general intelligence, and the prospect of robots taking people’s jobs, regulators within the EU and the US have been ramping up warnings against AI and emotion recognition. Emotion recognition is the try and discover an individual’s feelings or mind-set using AI evaluation of video, facial images, or audio recordings. 

But why is that this a top concern? Western regulators are particularly concerned about China’s use of the technology, and its potential to enable social control. And there’s also evidence that it simply doesn’t work properly. Tate Ryan-Mosley dissected the thorny questions across the technology in last week’s edition of The Technocrat, our weekly newsletter on tech policy.

Bits and Bytes

Meta is preparing to launch free code-generating software
A version of its latest LLaMA 2 language model that’s in a position to generate programming code will pose a stiff challenge to similar proprietary code-generating programs from rivals resembling OpenAI, Microsoft, and Google. The open-source program is named Code Llama, and its launch is imminent, in response to The Information. (The Information) 

OpenAI is testing GPT-4 for content moderation
Using the language model to moderate online content could really help alleviate the mental toll content moderation takes on humans. OpenAI says it’s seen some promising first results, although the tech doesn’t outperform highly trained humans. A variety of big, open questions remain, resembling whether the tool could be attuned to different cultures and pick up context and nuance. (OpenAI)

Google is working on an AI assistant that gives life advice
The generative AI tools could function as a life coach, offering up ideas, planning instructions, and tutoring suggestions. (The Recent York Times)

Two tech luminaries have quit their jobs to construct AI systems inspired by bees
Sakana, a brand new AI research lab, draws inspiration from the animal kingdom. Founded by two distinguished industry researchers and former Googlers, the corporate plans to make multiple smaller AI models that work together, the thought being that a “swarm” of programs could possibly be as powerful as a single large AI model. (Bloomberg)