Even just looking up a few of the quotes in the appendix made it seem less adversarial:
the U.S. and Russia should extend the highly successful Strategic Arms Reduction Treaty (New START) while negotiating a follow-on agreement.
This came up twice in the document.
in the vicinity:
2.
With the dissolution of the Intermediate-Range Nuclear Forces (INF) Treaty, New START remains the only bilateral nuclear arms control agreement between the United States and Russia. Its lapse at a time when China and North Korea (and potentially Iran) are increasing their nuclear capabilities and arsenals
#page=50
would send a harmful message to adversaries and partners alike while further undermining the global Treaty on the Non-Proliferation of Nuclear Weapons (NPT).116
1.
When gaming out the future of defense, the increasingly strained relations with China and the rising potential for conflict with Russia should remind the U.S. to consider diplomatic measures that adhere to the concept of statecraft versus the proliferation of war craft. Leveraging influence through economic, political, and social measures may be the most successful means of avoiding conflict while ensuring a stable, open, and transparent world order that allows democracies to thrive.
#page=22
(The url additions which jump around based on the pdf viewer's accounting of the page, are 6 more than the numbers in the lower right hand part of the pages displayed.)
Well, this is Committee on Armed Services - obviously the adversarial view of things is kind of a part of their job description... (Not that this isn't a problem, just pointing out that they are probably not the best place to look for a non-adversarial opinion).
The bolded parts definitely seem pretty grim. At least there was also
"Further, it will be incumbent upon the nations who use these technologies to set strong moral and ethical standards to protect the health and well-being of humankind. Advancements in AI, for example, will likely require a global compact in the vein of the Geneva Conventions, the Chemical Weapons Convention, and the Nuclear Non-Proliferation Treaty to establish guardrails and protect against a variety of factors, not the least of which is the infringement of personal liberty and freedoms."
So fingers crossed someone on this committee wouldn't deploy a couple million armed drones, or charge the NSA with sabotaging peoples' abilities to defend themselves from cyber-warfare.
Epistemic status: Quick and dirty. A surface level dive into a particular aspect of AI governance carried out over the course of one morning.
Context
The Future of Defense Task Force is a subcommittee of the U.S. House Committee on Armed Services. They have released a report, available here, and also as the first item on their latest news page. The task force is manned by an equal number of Republicans and Democrats. Though this seems a priori unlikely, it could both be the case that this report is unrepresentative of the political forces in the US Congress, and that this particular committee holds little power.
References to AI race dynamics in the report
Bold added by me.
The report cites: Michael Brown, Eric Chewning, Pavneet Singh, Preparing the United States for the Superpower Marathon with China, The Brookings Institution (April 2020) (online at https://www.brookings.edu/research/preparing-the-united-states-for-the-superpower-marathon-with-china/).
Brief commentary and discussion
The diagnosis in the report — that China will grow in power and technological acumen — seems basically accurate, though the recommendation that the USA should be willing to take more risk and go out with a bang seems more questionable. The language in the report is heavily adversarial, with an emphasis on winning an AI race, rather than with an emphasis on exploring and developing robust mechanisms to avoid Red Queen races.
The committee also seems to assume that AI scenarios will in the short term be multipolar, with the United States, China and Russia competing to develop their AI capabilities, while smaller nations also invest in asymmetric warfare. Europe's capabilities aren't considered at all. Crucially, AI is here seen as only one of many factors to consider in an engagement. Scenarios outside the Overton window, such as intelligence explosions, aren't considered.
I initially arrived at this report while researching this CSET question. Overall, I'd say that the language in this post is a signpost or warning sign of potentially more heated AI races to come.
It's unclear to me what levers there exist to make the US's approach less adversarial. Some brainstorming:
Best of luck to the folks working on AI governance.
Appendix: All interesting quotes I wrote down.
The report cites: Michael Brown, Eric Chewning, Pavneet Singh, Preparing the United States for the Superpower Marathon with China, The Brookings Institution (April 2020) (online at https://www.brookings.edu/research/preparing-the-united-states-for-the-superpower-marathon-with-china/).