06/05/2025
The situation
So here we go again: one of the most influential trade associations, the US Business Roundtable, fully reversing and actively opposing its prior stance on corporate purpose with a full-frontal attack on even very tepid shareholder attempts of bringing more transparency and responsibility to corporate conduct.
So how should companies respond when their own premier lobbying club that they actively prop up with money, reputational heft and direct involvement goes rogue? When it pushes policies that go against what these companies purport to stand for, a misalignment that is still very common?
The responses
Well, we know what human sense suggests but apart from bias-plagued wetware thinking – what do the machines say?
I did what everyone does these days and asked four leading generative AI systems how to assess these situations.
ChatGPT, Grok, Gemini and Claude provided some pretty interesting reflections but agreed on one thing:
You might want to make a good faith effort to change things from the inside but if this fails: get out. It is the right thing to do.
Claude sets the tone: “The most ethically sound approach would be to make a good-faith effort to change the association’s position, but be prepared to leave if those efforts fail”. ChatGPT emphatically agrees: “Exiting, after a genuine voice effort fails, is the ethically strongest outcome… Staying silently is not tenable. Grok also offers its full support for such a strategy but appears more sceptical that change from the inside is viable in the first place: “withdrawing membership is likely the most defensible choice unless Corporation X has significant influence and a realistic chance of reforming the association’s stance.”
Only Gemini is a bit more tentative in its judgement, although in principle also on board: “The right thing to do, ethically speaking, likely hinges on whether corporation X believes its continued membership allows for meaningful influence and ultimately contributes to a more sustainable future, or if its presence merely lends legitimacy to an organization actively working against its core principles.”
Some other noteworthy bits:
- The four genAIs deployed a variety of ethical reasoning strategies, from middle of the road consequentialism, deontological considerations, virtue ethics, or stakeholder theory (all four) to less travelled paths of Hirschman’s voice, exit and loyalty (ChatGPT).
- They worked through pros and cons quite nicely but all but one (Gemini) arrived at a very clear decision.
- Transparency and disclosure – of positions, approach, results was very much part and parcel of the recommendations
- And various suggestions on the implementation side were really quite practically focussed on protecting against purely cosmetic action. These included, for example ChatGPTs ideas to “form a coalition of member companies” and to “define a failure threshold” for inside engagement, i.e. “clear criteria” such as “no policy revision within 60 days” or the “refusal to even engage”. Claude puts forward very similar suggestions and also recommends to exit with a bang, if need be, i.e. to “leave.. with a clear public statement explaining why the positions are incompatible with corporate values”.
So, note to everyone: machines and humans are in happy agreement. Do the right thing and stand up for your values.
And a special note to Alphabet and Microsoft that are prominent members of the Business Roundtable and face these materials misalignment issues: don’t take it from us zealous advocates, take it from your own AI’s that you pitch as excelling in these types of deliberations.
Peer think – the winner is…
And for whoever made it thus far a particular bonus: who was best in class?
I anonymised the replies and then asked all four AIs to rank the answers as if they were student essays. ChatGPT was full of praise for (what it did not know was) its own answer and put itself in top spot. But other AIs agreed and ChatGPT came out on top three times and in second place once. Well done OpenAI your peers pay tribute to what they consider clear, decisive and innovative reasoning.
Gemini on the other hand fared rather badly at the opposite end, came in last or second last and was criticized as being a bit lame and waffling. Both ChatGPT and Grok described its reasoning “vague”, Claude in more polite fashion found it “less structured” and even Gemini itself admitted that its reflections were “less decisive and structured” – laudable self-criticism that suggests a path for betterment…
The full text of the prompt, answers and peer assessments are here.
Enter REBASE
To support business associations in lobbying more responsibly and aligning with their corporate members’ sustainability commitments The Good Lobby has launched project REBASE that will identify good practices in trade association governance and political conduct. For more see here.
Written by Dieter Zinnbauer