Time for Philosophy’s COI Policies to Grow Up (guest post)
“Philosophy’s broadening impact has generated increasing interest from industries looking to partner with academic philosophers. Ties are now especially common between philosophers and technology companies producing AI products. These changes call for discipline-wide reflection on the norms of research integrity.”
That is an excerpt from an open letter calling “for philosophy journals to strengthen or implement new policies requiring author disclosure of relevant industry ties and conflicts of interest (COI).”
The letter was written by Cailin O’Connor (UC Irvine) and Craig Callender (UC San Diego), who explain their reasoning for it in the following guest post. “Norms surrounding conflict disclosure (financial and non-financial) are basically non-existent” in philosophy, they say, and “improved disclosure is one easy first step” we can take to “protect the integrity of our research.”
Time for Philosophy’s COI Policies to Grow Up
by Cailin O’Connor & Craig Callender
Both of us have been working on industry mis- and dis-information in science. By the mid-twentieth century it was clear cigarettes caused lung cancer, but the tobacco industry developed a suite of techniques to influence scientific research on a range of topics, which helped protect their products for decades, allowing them to rack up profits even while millions died. Fossil fuels, agrochemicals, pharmaceuticals, sugar and many other industries use more or less the same techniques to protect a range of products, sometimes with disastrous consequences for public health.
The “tobacco strategy” of influencing science often involves techniques that work within the norms of academia, and that may not tip off academics to the role they play in industry influence campaigns. Businesses need not directly fund research in order to skew the results of inquiry in their favor. For example, industry can support just those areas of inquiry that yield desired findings—for instance by emphasizing the role of social media users in the spread of misinformation, rather than platforms, or by highlighting the role exercise plays in health, rather than soda.
We recently started circulating an open letter aimed at improving conflict-of-interest disclosure policies in philosophy journals. We encourage you to sign and share it. Why? It is past time for philosophy to start worrying about the tobacco strategy in our own discipline, and to take actions to protect the integrity of our research. Improved disclosure is one easy first step (though likely not the last we’ll need.)
In the early 2000’s some universities stopped accepting tobacco money. Some journals stopping reviewing papers funded by tobacco. Top medical journals, like JAMA and The Lancet, tightened their conflict of interest policies, as did Science and Nature. Obama’s Sunshine Act created a public registry cataloging gifts to medical doctors. In fields like biomedicine it became a norm to disclose funding at societies, journals, academic units, and in individual talks.
None of these policies are perfect, of course. But comparing them to those typical in our own field, we notice a dramatic difference. Philosophy’s conflict-of-interest policies are very far from the state-of-the-art. Norms surrounding conflict disclosure (financial and non-financial) are basically non-existent.
No doubt a big reason for this is that historically not much private research money flowed into philosophy. When Descartes shut himself in a room with an oven and had his visions, he wasn’t paid by Big Masonry Heater. And philosophers like to think of ourselves as unaffected by external forces. So even when more private money started coming into our field—Charles Koch Foundation, Andrew Mellon Foundation, Templeton Foundation, Open Philanthropy—we never really developed policies and norms like those in other fields.
Now, Big Tech is sponsoring a lot research in AI ethics and adjacent fields. Money is coming in through labs, centers, fellowships, and consulting. Some fear that AI ethics has been captured by industry. This may or may not be right, but we know industry has a vested interest in the outcomes of this research. Our field needs to improve policies designed to protect research integrity in the face of potential threats like this. Hence our open letter. The problem is more general than tech and AI, but its threat helps us see that it’s time the field’s conflict-of-interest policies grew up.
As of the publication of this post, the letter has around 170 signatories. You can read and sign it here.
Just a note to say that the open letter has been shared with editors of The Journal of Philosophy, and we’ll consider it ahead of reopening submissions next month. Thank you!
Fantastic!
“Improved disclosure is one easy first step (though likely not the last we’ll need.)”
Full disclosure of my own: I do contract work for an AI company, so I’m squarely in the population this letter has in view. I have no objection to the disclosure norm itself — I already file a COI disclosure as a university employee and will be adding the AI work when I next update it. While I’ve never published in philosophy journals on matters related to AI, were I to do so, I’d be happy for readers to know about my industry connection.
My question is about the parenthetical. The post’s historical examples suggest what later steps might look like. Some are more disclosure–e.g., creating a registry of gifts, which I imagine wouldn’t have many entries in philosophy, but couldn’t hurt–but some are quite different: journals declining to review industry-funded papers or institutions refusing industry money. Are those the sorts of measures the authors have in mind for philosophy? A norm favoring COI disclosures at the end of articles strikes me as innocuous; a norm against reviewing industry-connected work would be something else entirely.
Hi Daniel, We’re just advocating for transparency, not disassociation. In some special cases I personally think disassociation is warranted, as in public health and tobacco. Regarding anything more than transparency, note that there are many other policies possible that go further than transparency but fall short of disassociation. At UC, for example, we can take tobacco money, but taking it triggers a special review process. But one can also reform gift review committees, add an internal ethics-like IRB to the process, and implement other similar types of reforms to create better guardrails.
Thanks!
This seems like a good idea, even as someone who’s also “squarely in the population this letter has in view.” Letter signed!
My main concern is to not make reporting requirements overly burdensome, unless you’re actively trying to dissuade submissions from these folks.
For instance, the US National Science Foundation now asks for info about collaborations, affiliations, associations, etc. That also seems to be a good idea, except that each one of those things is a new entry on an NSF form, which isn’t as simple as just listing the organization.
Also, some of those things are under-specified. E.g., if I’m invited to an AI conference, and travel expenses are covered, is this an affiliation or association to declare?
I’d think it wouldn’t be, if there’s no other association or gifts or payment, beyond expense reimbursement. Anyway, it’s probably impossible for me to figure out who exactly was sponsoring all my conference trips in the last 20 years. I know that professional journalists, as an example, see it differently; even travel support muddies the water too much for them. But I’m not aware of an academic discipline that also takes this view.
A wrinkle: what if the conference sponsor was someone high profile and controversial, like Elon Musk? If the answer changes, why would it? And does it matter that just about everything I’ve written related to him was critical; is that something that needs to be investigated for each case, to determine if it’s reportable?
As usual, the devil’s in the details, and I’d suggest not going overboard with these disclosures or expanding (too much) what is considered to be a conflict of interest given disciplinary norms.
For example, someone can also make a similar argument for disclosure of religious and political affiliations, incl. voting record. There’s value in this, in understanding those pre-existing commitments to identify possible motivated reasoning, which seems to be the same danger we want to guard against re: COI.
But we also need to understand what we might be losing with increasing disclosures, in order to strike the right balance. E.g., are there any red lines, such as disclosing one’s income and assets? Maybe there’s a philosophy paper in all this…
Good luck with your open letter!
I don’t think the idea that disclosing is onerous should stop us from surfacing potential attempts by industry to influence our discipline. (And travel gifts fall squarely under typical policies for disclosure, as professors in CA we are required to disclose when we accept private funds in just such cases.) External influence from industry and similar groups is also very different from the normals sorts of commitments -political and religious- that might sway research. This is because when industry is getting involved in academia that is often because they are trying to protect a dangerous product. Ie, they are systematically pushing research away from protecting public health.
Hi Cailin, you said:
I think you’re pointing out a good example of why this can be confusing. As I understand it, no, we California professors are not required to disclose travel funding, if we’re not “designated employees” (i.e., who make decisions for the university) and if the third-party funded trip isn’t part of a research grant. There are other nuances and exceptions, such as if the funder is a US gov’t agency or non-profit.
Also, I wasn’t saying that any requirement to report is too onerous, but it’s possible that some reporting schemes (not necessarily yours) can be too onerous and become real barriers. I was just curious about where those red lines might be.
Finally, I understand that you feel that industry relationships are different from other kinds of relationships, but that’s an argument you’re making, which might not be obvious to everyone. Or at least a plausible argument could be made for disclosing other relationships and personal info.
For instance, I can think of many examples of organized religion and political parties are also trying to protect a dangerous thing, i.e., they are systematically pushing research away from protecting us from the thing. So, the difference between industry influence and other influences might not be as clear as you think.
Now, if you’re saying that your open letter will push for reporting of all travel funding, that that’s precisely an example of an overly burdensome requirement, at least for those of us who travel a lot…
Hi Patrick,
I’ll just add that on travel, that was of course one of the most famous perks bestowed on doctors from Big Pharma. Industry paid for junkets to Bahamas, Paris, etc. Then in the early 2000s professional societies, journals, and universities started policies to deal with that.
What to disclose is a tricky issue. I don’t think there is a red line. Rather, I see fields continuously grappling with it given the evolving nature of the costs and benefits of imposing/not imposing the burden. If AI ethics conferences were all in Hawaii, lavish, not rigorous, and only open to those who never advocated regulating AI, that would start to look like a perk and raise eyebrows. Anyway, without knowing the answers in advance, this is the sort of discussion we hope to prompt.
Thanks, Craig. Yes, paid travel can be a COI for certain industries, even beyond journalism.
The tricky part is that for the vast majority of academics, traveling to conferences and so on is a part of our expected work and not a presumptive COI. But, yes, possible COIs can arise even for “normal” academics, e.g., if a quid pro quo is expected or implied from a conference sponsor.
I’m not at all an expert in any of this, so again best of luck with this work!
Also, Cailin, in Science’s COI policy you had linked to in the open letter, note that they require disclosure of personal beliefs where relevant:
And neither Science’s nor Nature’s COI policy mentions travel support…
This idea of disclosing personal beliefs and personal relationships seems very tricky at the stage of authorship. For peer reviewers and anyone making judgments of quality on manuscripts that advance positions that may be contrary to deeply held personal beliefs is where the attention is merited.
While I think at this stage any such person would recuse themselves without even having to disclose those beliefs, I would hope the editors would be able to identify that degree of bias in the submitted review and send the manuscript to one other reviewer if they believe the review is compromised.
If the disclosure of personal relationships is expected (I can see some of them being particularly worrisome in terms of conflict), I am not sure these should be made public by default, although perhaps they should be provided by request to readers … I know this is an issue in terms of think-tanks, journalism, pundits, etc. I would hope the higher evidentiary standards in scholarly journals would take care of these on their own merits.
On the travel: If I am being invited to attend a conference or gathering or summit or organizing/strategic meeting, all expenses paid, and my manuscript or other publication related activity overlaps, then I think this needs to be reported. It is the same as getting paid.This does not mean the scholarship should be assumed to be biased as a result (this is a really important point that letter should probably make more clearly, if I missed it), but it is useful contextual information. In fact, I sometimes think that access to such spaces sometimes strengthens one’s standpoint as a scholar on certain subjects. I am thinking anyone working on institutional or political corruption, or academic freedom, might come to the table with better awareness than others. Being party to an event does not mean one is party to its outcome and certainly not that one’s scholarship will be partial to the outcome.
One option is something like a ‘register of interests’. Perhaps my subsidised trips to snail farms so far don’t seem worth disclosing; but if I suddenly start writing papers on philosophy of, say, taxation, sponsored by Big Snail, I can then list all my trips on such a register, and then simply refer the interested reader to them. This also allows people to see how someone’s potential conflicts of interest have evolved over time without having to go back and update old papers. I’m not, admittedly, aware of any discipline with comparable norms.
Don’t worry: when it comes to AI, many will rush to do the apologetics for free.
Honest question: should this COI policy (which I think makes sense if it doesn’t overreach) also apply to reviewers and editors?
As gatekeepers for the journal, they seem to be the most influential to whether an article makes it into their journal. I can imagine a pragmatic reason to exclude them for now, but is there a principled reason?
A brief, immediate reaction:
It seems like this would be expected and considered through the process of selection in Editorial Boards, and then the addition of a link for public disclosures would be acceptable. And for reviewers, it seems like when reviewers accept the assignment, they are asked to disclose any possible conflicts of interest relative to the area of review. Write into policy. The manuscript authors won’t be able to know, but the journal board is responsible for making the judgment. Of course, the reviewer can always lie, but that’s at their own future reputational peril. It would probably include an extra step when inviting scholars to review.
Our authority is still mostly based on the conviction that we are committed to the evaluation of knowledge.
Frankly, I assumed all of this was already implied for manuscripts and assumed it was considered when composing Editorial Boards, even if not public.
There are researchers who sign non-disclosure agreements with corporate parties. For example, a researcher may be paid to work on the evaluation of the environmental impact of launching rockets. How do you think this can be handled?