Use reached a responsibility boundary before measurement
Hank Green's production episode and Altman's family-audio pitch provide no adoption measure; they expose different social boundaries between delegated assistance and retained responsibility.
Artificial Intelligence··Morning
A production trace made the workflow visible
The dispute around Hank Green began when viewers noticed a phrase in a Complexly video script that appeared to be residue from a chatbot response. Green explained himself in a since-deleted post on X and in the nerdfighters community on Reddit. By his account, he used ChatGPT to research that script and to locate papers and other sources, while the words and interpretations remained his own. Even so, he acknowledged the trust problem created by the workflow, said he had disappointed many people and described himself as mortified. He called the reward he gets from interacting with large language models unhealthy for himself and for the world. His main channel has 3.2 million subscribers. Green said he would pause his channels or publish less often and reduce video production. The episode supplies no product-usage rate; it is one case in which unseen research assistance became a public dispute about production and responsibility after a trace appeared in the final script.[1]
The proposed family programme tested another boundary
Sam Altman's shared use case came not from a production back office but from the space of attention and conversation inside a family. The proposal connects family calendars to ChatGPT Work, describes the children's interests, and asks the system to prepare a daily programme for the school run. The example combines one child's football game that afternoon, another's approaching birthday, and several items from the news. Responses travelled farther than the pitch. Animator Alex Hirsch's reply suggesting direct conversation with children drew 9,000 reposts and 122,000 likes, while Altman's post had 300 reposts and 9,600 likes. That comparison measures engagement with two particular posts; it does not show whether parents use the product or continue using it. The report also contains no company figure for the prevalence of this scenario. The tension formed less around whether the system can generate personalised material than around how much of the work of turning family context into conversation should be delegated to it.[2]
The episodes show two different relational conditions for adoption
In Green's case, the tool had already entered a production workflow; the boundary appeared in a dispute over how that assistance was represented in the final script and who remained accountable to the audience. In Altman's example, use was still a proposal; the boundary appeared in the question of whether converting personalised family information into an automated programme displaces attention between parent and child. The first produced a retrospective explanation and a decision to reduce output. The second produced a proposed product scenario and a visible response to it. That distinction matters. Collapsing them into one measure of acceptance or rejection would therefore be misleading. Together, however, they show that socially adopting a technically possible use can depend on conditions that keep human responsibility visible. Green's account is one creator's experience, while the replies to Altman are social-media engagement; neither is representative user research. Their shared thread is that users may have to specify which relationship and accountability they retain before the question of what work the tool can perform settles adoption.[1], [2]
Related columns
For more information on this topic, you can read the related columns.
