The London School of Economics and Political Science (LSE) and the University of Toronto have launched new initiatives to support and focus philosophical research on various aspects of artificial intelligence (AI).
The Department of Philosophy, Logic and Scientific Method at LSE has received a $5.3 million donation to launch a new research group—The Global Priorities Research Group (GPRG)—dedicated to “determining which problems matter most and what should be done in trying to realize the best outcomes for humanity and the wider world.”
It’s primary emphasis will be “philosophical research that helps us navigate the largest-scale challenges associated with the impacts of transformative AI.”
Funded by Good Ventures, a philanthropic foundation, the GPRG “will also consider how we should compare and weight the interests of different kinds of beings—be they human, animal, or perhaps even artificial—as well as pursuing foundational research on issues related to value and rational choice with the potential to reorient our priorities across a wide range of issues.”
The group will be led by philosophers Andreas L. Mogensen, who recently joined LSE from the University of Oxford, and Teruji Thomas (Oxford, LSE), and will be based at LSE’s Centre for Philosophy of Natural and Social Science.
Meanwhile, the Artificial Consciousness Initiative is a new project from the University of Toronto Department of Philosophy that “aims to advance foundational research connecting artificial intelligence and consciousness.”
The project’s director, philosopher Andrew Y. Lee, writes:
The initiative’s research agenda focuses on questions about the nature, structure, and indicators of artificial consciousness; the valence and character of potential artificial experiences; the welfare and moral status of artificial systems; and the ethical and safety risks raised by increasingly capable AI. The initiative does not presuppose that current artificial systems are conscious or that future ones will be. But it starts from the view that these questions are worth taking seriously.
The core team of the Artificial Consciousness Initiative includes Karina Vold, Christian Tarsney, and Gwen Bradford.
The initiative’s aims include basic research, partnerships with other institutions, and contributing to public and policy discussions related to AI consciousness. The project is supported by a large grant from Longview Philanthropy.
Both initiatives are currently hiring postdocs.
This is excellent news. I know that I am deeply concerned every single day when I look at philjobs and see so few offerings for people working on philosophical issues related to AI.
“will also consider how we should compare and weight the interests of different kinds of beings—be they human, animal, or perhaps even artificial”
This is a trojan horse. The encroachment of weighing the ‘interests’ of [corporate owned] artificial beings, against the wellbeing of humans and animals, and at a time like this, is insidious.
My thoughts exactly.
I have no view on these specific initiatives, but it seems quite plausible to me that there will someday be artificial beings that are not controlled by corporations. If it’s even plausible that this could happen, it’s quite important to determine how to weight the interests of such beings, to be prepared for if they do exist. Is that assumption totally implausible to you?
For what it’s worth, I hope that seriously considering the possible interests of artificial beings will lead to a radical reconsideration of our relationship with animals. We need to decenter humans here, and recenter all beings with interests (which may or may not include AI).
Yes, it’s very common in the AI space to argue from a hypothetical future technology that does not resemble current AI to certain conclusions, and then apply those conclusions to the current technology. Philosophers should be careful not to launder tech talking points.
They should also be careful not to launder lazy clichés found on social media. Max makes a perfectly reasonable point.
Great comment, Max.
Just a thought: What the corporations are doing with AI seem to be self-defeating though also in this regard. If they do create artificial beings, chance is those being will have a moral status like ours, at least as moral patients, but likely as moral agents as well. If so, unless we now also sign up to slavery, the precious products of the corporations will become autonomous agents and the company will no longer have control over them. Is it worth running this risk?
What is insidious about examining whether artificial beings can have interests and what weight those interests might have? What is the hidden thing in the “Trojan horse”? Are Andreas Mogensen and Teruhi Thomas up to something insidious?
What implications does being owned by a corporation have for a being’s interests? Are there implications for the interests of factory-farmed pigs from their being owned by corporations?
I take it there are implied claims in your posts that add content to the accusation, but (at least to me) it’s not clear what they are. It would be helpful if you articulated them.
I’m not quite as cynical as “Despondent” (though their comment got a wry grin from me). But I am cynical enough that I would really like it if you’d say a bit more about the funders when announcing the launch of things like this. As one should guess from the name, Longview are an openly longtermist effective altruist organization and its funders, to the extent one can find that information, include a crypto mogul and a former Jane Street trader. Good Ventures is almost entirely funded by Dustin Moskovitz and Cari Tuna, tech billionaires and card carrying effective altruists.
So I am very worried that at worst these are attempts to ethics-wash what AI companies do. And even at best, we need to be aware of the ways that funders like Good Ventures and Longview will shape the debate in ways that foreground certain issues– skynet killing us all to be blunt about it– and ignore others like implicit bias in AI, intellectual property theft, deskilling of labor, environmental damage, responsibility sinks for targeting in war, and exacerbating what I’d describe as anti-humanist approaches to education just to name a few. None of that is to dismiss out of hand anything these initiatives might produce, but if analytic philosophers have any pretensions to critical thinking they need to ask the “Who’s paying for this and what do they want for their money?” questions far more often than they do.
I agree that it’s important to note where the money is coming from, and I am also worried about some of the issues you note here, Sam. But I would also add that in addition to shaping the debate in terms of which issues are given consideration, it is also important to keep an eye on which approaches are adopted in considering issues.
There are very different approaches one can take when “comparing and weighting” different interests. For example, do we compare the interests of currently existing interest-holders only? Include also the interests of those who do not yet exist? How do we weight the interests of existing versus merely possible interest-holders? Effective altruists take positions on these questions that are not universally shared. Indeed, there are respectable (even if one doesn’t agree with them) positions that hold that comparing and weighting are problematic approaches to begin with. Will these funders be willing to fund work that challenges core assumptions of the EA movement? What shape will the debate take given the background commitments of the funders? What measures are in place to make sure that particular approaches are not excluded from the outset?
I’m always puzzled in these discussions by why people treat effective altruists as an inherently insidious force, when they never raised any similar objections about (e.g.) Mellon funding social justice priorities.
Different people have sincere prioritization disagreements, and I think they’re better argued on their merits than through silencing accusations of “ethics-washing” or the like. (I appreciate that you wrote, “None of that is to dismiss out of hand anything these initiatives might produce,” but I’m not sure what other role these sorts of accusations play if they’re not attempts to delegitimize the work of people who “foreground certain issues” that you apparently don’t want to see discussed, and to discourage other people from even considering their arguments. Dismissive “skynet” rhetoric reinforces this sense of well-poisoning.)
If we’re going to have a discussion about the role of external funding in the discipline (and whether it’s a good or a bad thing), it seems worth trying to pursue that in a more content-neutral way rather than selectively demonizing those whose sincere, reasonable and defensible priorities you simply happen to disagree with.
As I wrote in (When) Does Funding Corrupt?, I think the key question to ask is whether we should expect external funding to increase or decrease “distortion” relative to the ideal allocation of academic attention. I think it’s important to recognize that internal social dynamics, fads, and biases have a far greater distorting effect on the discipline than external funding does, and the latter can plausibly serve as a helpful corrective by supporting work that falls outside of the usual academic monoculture. For example, I think Templeton funding (back when it supported philosophy more) was plausibly good for this reason of intellectual diversification, even though I don’t personally share their starting points.
I actually disagree with your closing claim that philosophers should focus more attention on ad hominem questions about people’s motives rather than assessing arguments on their merits. As I concluded my post: “I worry about our more ideological colleagues pouncing on any excuse to dismiss or delegitimize disagreement. In my experience, it’s already rare for philosophers to be willing to invest in serious cross-camp engagement. The last thing we need is more excuses to avoid hearing each other out.” But if you are going to go looking for biases, at least take care not to do it selectively.
I’m fine if people want to talk about malicious AI, and while we’re tut tutting with Critical Thinking, implying that I somehow want to censor them is pretty obvious strawmanning. What I don’t want is for this to consume 90% of the debate while other issues I judge a lot more important are relegated to afterthoughts if they’re discussed at all.
It’s also no mark of wisdom to shut down any discussion of these things by tarring them as “ad hominem” arguments. Is it wisdom to not treat arguments I read in the “Financial Times” about how the Iran war is going very badly a bit differently than those from the White House press secretary that it’s going just swimmingly? Am I making a basic error when I treat a Kia salesman’s arguments about the relative merits of Kias vs Toyotas with more skepticism than those of my cousin’s husband who’s been an independent mechanic for ten years? Intelligence can sometimes look like naïveté but it’s still a very different thing than willful ignorance. (For what it’s worth I’ll cop to a bit of a false dichotomy on that last one since Hunter says Kias have gotten a lot better in the last ten years and are now a decent car for the money).
I’d welcome more disciplinary attention to the question of what issues are “more important”; that’s precisely the sort of question global priorities research aims to address. I don’t think there’s much risk of any one topic dominating (anything like) 90% of discussion in the field. But like I said, it’d be worth thinking more about what the ideal allocation of academic attention would look like, and what factors are most distorting our collective efforts away from that ideal.
I’m just dubious of people selectively attributing bias to those they disagree with, without any attempt at a more systematic examination of the problem. Philosophers rarely know enough about the actors in question — a point my linked post made with regard to overlooked divisions in AI politics, and reinforced by your backwards assumption that EAs like Moskovitz want to “ethics-wash what AI companies do”. Do you think that EAs are on the side opposed to regulating AI labs? How do you take your “Kia salesman” analogy to apply to the present case? Are you assuming that Moskovitz and Tuna personally profit from AI labs?
Outgroup homogeneity effect badly distorts the judgment of people who see “tech billionaire” and little else. Scott Alexander’s AI Safety Is Not A Corporate Plot could be helpful background reading.
“implicit bias in AI, intellectual property theft, deskilling of labor, environmental damage, responsibility sinks for targeting in war”
These are not exactly under-discussed issues. It’s like 90% of what’s already going on in AI ethics. I’d wager the majority of AI&Philosophy jobs are going to people who hold the sort of views you consider acceptable rather than to EAs. Rest assured, those two modest initiatives are unlikely to alter the academic status quo.
I’ve never quite understood philosophers being so worried about the sources of funding for philosophy… hardly anyone funds philosophy, and as long as they are not making the advancement of any viewpoint a precondition of the funding, then as I see it failing to take it is like being in a desert when there is only one hospital within 100 miles, and your leg needs fixed, but you refuse to go because some of that hospital’s investments might be mixed up with the Saudis or Bezos or pick your poison.
Thanks, Sam, for bringing up this important issue regarding the sources of research funding. Many readers may not have the time or background knowledge to look into this. And many may not even think it is a salient issue. Philosophers are way behind with respect to this important norm in scientific and medical research. In these disciplines, when asked to dislose conflicts of interest it is expected that one disclose PERCEIVED conflicts of interest. Not disclosing can become ground for the retraction of publications afterwards. I have received a number of grants from various funding bodies and I always disclose who is funding my research. Such information needs to be flagged – the rest of us should not have to embark upon detective work just to find out who is behind the funding.
The post discloses the funding sources in perhaps more detail than you’d get in a journal publication. No more detective work is required than when you disclose your funding. But if people took the time then instead of discovering nefarious origins they would learn, for instance, what Good Ventures and Longview Philanthropy actually do, or who Dustin Moskovitz is and all the things he’s done, and then they might wonder why Sam Duncan sounds so conspiratorial.
Some readers of this post might be interested in a BIOPOLITICAL PHILOSOPHY post that Mich Ciurria wrote on the topic. You will find Mich’s post here: From Bioethics to AI Ethics: A Genealogy of Philosophical Sell-Outs – BIOPOLITICAL PHILOSOPHY
In various contexts, I’ve written about the neoliberal corporatization of philosophy through the emergence and institutionalization of bioethics, including in my article “Note for a Feminist Abolitionist Genealogy of Bioethics,” which appears in the current issue of Feminist Philosophy Quarterly. You will find my article here: Notes for a Feminist Abolitionist Genealogy of Bioethics | Feminist Philosophy Quarterly
It’s incredibly sad and frustrating that so many feminist and other oppositional/ progressive philosophers have (naively?) succumbed to these institutionalized mechanisms of neoliberalism and, in doing so, have effectively ruined the careers (and indeed lives) of an increasing number of marginalized philosophers.
Here is the ad for the jobs at the LSE (two four-year postdocs):
https://jobs.lse.ac.uk/Vacancies/W/5138/0/475866/15539/research-officers-for-the-global-priorities-research-group-gprg
Two notes from the ad: “The ability to produce research related to…transformative artificial intelligence is desirable but not essential”. And, when it comes to successful candidates, “At least 50% of their time will be spent on topics directly relevant to the group’s research agenda, which they will help to set.”
Fantastic news, godspeed to all involved
Inspiring to see talented researchers focusing on these important questions!
Sometimes I wish that the discourse on funding like this could stay closer to the middle ground between “this is pure evil” and “any money for philosophy is great news.”
I’ve worked for and with institutes like this in the past, and I fully plan to visit the LSE folks soon. I’ve also criticized their funders publicly. You can do both. They’re actually remarkably open to it, and there is progress to be made on important philosophical issues in the process.