AI code review (Claude, maybe Codex)

[Caleb James DeLisle]

On 08/04/2026 09:57, Jean Thomas wrote:
>
>
> Le 08/04/2026 à 09:04, Thibaut via openwrt-devel a écrit :
>>> Le 8 avr. 2026 à 00:51, Hauke Mehrtens<hauke at hauke-m.de> a écrit :
>>>
>>> On 4/7/26 08:35, Thibaut wrote:
>>>>> Le 7 avr. 2026 à 05:57, David Lang<david at lang.hm> a écrit :
>>>>>
>>>>> Hauke Mehrtens wrote:
>>>>>
>>>>>> On 4/6/26 19:33, JP wrote:
>>>>>>
>>>>>>>   - these platforms are subsidised (in the extreme) by (provably 
>>>>>>> society-damaging) VC-funds; any attempt at building 
>>>>>>> infrastructure upon this without significant 
>>>>>>> review/planning/estimation strikes me as potentially high risk
>>>>>> Isn't this good? OpenWrt can profit from these VC-funds.
>>>>> exactly, where is the project risk? there is no talk of 
>>>>> eliminateing all manual review (even if "AI" approves it, that 
>>>>> doesn't mean that it's the right thing for real hardware, or the 
>>>>> project overall)
>>>> Seeing how a lot of people (myself included) hate to have to deal 
>>>> with a chatbot any time they're trying to reach a human, I’d say 
>>>> the risk here is to put off contributors if all they get (or the 
>>>> first thing they get) is a chatbot review.
>>> Currently we put off contributors by not reacting to their PR at 
>>> all. There are many PRs which did not get any comment at all. On 
>>> many people only add the first comment after more than a week, then 
>>> the contributor is often not interested any more and this comment 
>>> was useless.
>>> The CI is running on all PRs, it complains about too long commit 
>>> titles and similar things.
>> Don’t get me wrong, I totally agree with all that assessment. What 
>> I’m suggesting however is that the issues will only compound: unless 
>> you plan on giving commit access to the chatbots (or let committers 
>> blindly commit chatbot-reviewed submissions - and they would still 
>> have to find time and interest to do that, two things which are often 
>> missing it seems), I suspect we’ll soon see a crop of bitrotting PR 
>> where the only interaction with the submitter will be that of a 
>> chatbot. And I don’t think this is going to improve their frustration 
>> or the project state and standing.
>>
>> tl;dr: I don’t think chatbots can fill in for the humans in the loop, 
>> but maybe we disagree on that topic.
>>
>> My suggestion would be at the very least to not make the chatbot 
>> review automatic: at least if there is a human chatbot-request in the 
>> PR history, for the submitter it suggests that*someone* is paying 
>> attention.
>
>
> I agree with you, a lot of recent studies show that involving AI only 
> adds more work for the humans in the loop, and do not decrease their 
> workload as expected.
>
> My experience with the few PRs I opened on OpenWrt was indeed quite 
> frustrating, as I had to wait and ping maintainers for weeks in order 
> to make some progress and have them merged, but having an automated 
> response without any further interactions would not have made it better.


I may be too old to fully appreciate this AI stuff, but I'm also too old 
to have any strong political opinions anymore.

My only comment would be that I think if someone invokes the bot, they 
should really make a point of coming back and reviewing the response - 
because these bots are known for giving nit-picky or even outright bad 
notes, and the person making the PR might think the bot's reply is a 
checklist for getting merged. And they'd be extremely frustrated if they 
addressed all of the notes only to have the PR still languish.

To face off this risk, it might even be worth having the bot say at the 
very top: "Dear $person_who_invoked_bot, here are my notes on this PR, 
please review them and tell $person_who_opened_pr if there is anything 
you would like changed."


Thanks,

Caleb


>
> I do not say that the solution is easy: OpenWrt faces the same issues 
> as a lot of open source projects, with only a small team of active 
> members and a lot of work to do. But this is a human and community 
> issue, and adding a new tool will not make it go away.
>
>>
>>>> Besides, (and perhaps more importantly) this doesn’t really address 
>>>> the fact that using chatbots is actively contributing to setting 
>>>> the planet on fire, if that’s something that matters to the project.
>>> Processing this mail also consumed a lot of energy.
>> Completely different orders of magnitude, and I’m sure you know that 
>> very well.
>> Last I checked, you don’t have to pay by the byte to read your mail.
>>
>>> It will be scanned by at least 100 spam filters and the AI companies 
>>> will use it to train their system.
>> *sigh*. The AI training is parasitism, it shouldn’t be taken as part 
>> of the « normal energy usage » of a mail system. And I’m sure you 
>> know that too. It looks like you’re saying is « energy is wasted 
>> anyway so it’s fine to waste more ». Is this really the best 
>> counter-argument you could come up with?
>>
>> The energy footprint of the project is already pretty bad as it is, 
>> with e.g. the builders building the same bytes forever and ever (I 
>> guess as long as there are people willing to foot the bill it’s ok), 
>> but the carefree attitude is quite disheartening. It promotes the 
>> idea that it’s ok not to care about these issues, and that’s exactly 
>> what will eventually incinerate the planet. But maybe nobody cares 
>> about that and I’m the outlier here, so I’ll shut up.
>>
>> T.
>
> You are not an outlier.
>
> AI should be seen as what it is: a tool built on the work of countless 
> humans and open source projects without any regards for their 
> licences, destroying the planet by using always more water and energy 
> even if these are already scarce resources, and requiring the work of 
> underpaid workers to train their model.
>
> Jean
>
>