The ethics of doom-trolling
In June 2026, Anthropic put up a paper When AI Builds Itself. It suggests that AI is about to design and create newer versions of itself without human input. This is frightening as it brings to mind scenes of robots replicating themselves and taking over humanity - not helped by the virus looking animation the paper comes with. This is the second in a series of three blogs looking at this paper through the three lenses - data & AI literacy, data & AI ethics and data communication.
Let’s look at the ethics of these types of papers.
Doom-trolling
I absolutely love this term coined by Cal Newport in a piece titled ‘Dear AI Companies Stop The Doom Trolling’ for the New York times. Doom-trolling to me is the tendency for the big players in the AI hype machine to tap into our worst fears built up over hundreds of thousands of sci-fi movies & stories where robots take over the world, and we humans become subservient slaves to the machine - barely surviving - when in actual fact this is not what is happening.
Why this matters
With every paper, tweet, podcast that does this, everyday ordinary people are feeling scared, overwhelmed and they are losing hope for the future.
Keeping us scared reduces our ability to ask questions or to dig deeper into what’s really going on so that we can make good decisions about AI. Who knows what we’re missing with this constant onslaught of doom-trolling.
But aren’t they just warning us?
Some people argue that the AI companies are actually doing us a service by highlighting the ‘risks to humanity’. They see the AI companies as being open & honest as opposed to cigarette & fossil fuel companies who knew the damage they were inflicting and spent bucket loads of money to keep it hidden.
But I disagree. I don’t believe that they are being open & honest with us for the simple reason that their claims often don’t stack up [I show an example of this in the blog post what aren't they telling us]. In fact, when you look deeper into their claims of sentience, consciousness, singularity to super intelligence etc. there is either a rational explanation, a lot of work humans do [that barely gets a mention!] or it’s just musings on possible fictional futures.
So, while I can only speculate on their motives as to why they are going all in on the doom-trolling - based on their actions, I don’t think it’s for the good of humanity.
What should they be doing?
In my mind, if they were truly worried about doing the right thing by us, we would see them:
Giving us the whole picture including
the total costs and time (& that includes the training of models)
the human effort behind the scenes
the proportion that comes from LLM’s vs smart wrap-around harnesses vs sheer compute power
Designing a system that isn’t based on optimising engagement
Pricing it so that true costs are reflected
Investing in as much as understanding how the models work as they do in training the models so that we can start putting in place better controls and provide full transparency
Lobbying together for external experts
to monitor the safety of AI
to regulate AI usage & development (without making it too onerous for competition to enter)
to determine a ceiling - just like we did with nuclear power & cloning
Doing these will allow us to make calm, informed decisions about how we want AI to be a part of our futures. Instead, these companies seem to go to great lengths to keep us scared - leaving us unable to even think about questioning what’s going on - let alone ensuring that AI works for us.
The Challenge
This is a hard one, as I don’t expect the AI doom-trolling to go away anytime soon, but whenever it appears - take a deep breath and think about why they might be saying this - are they falling behind in benchmarks, are other companies ramping up their claims, are they about to start another round of funding or try for an IPO? It may well be that there is something else motivating them rather than any real concern for humanity.
