iiinova — where your digital self keeps evolving

Boris Lenhard @borislenhard.bsky.social · 04/10/2026

FOXP2 has a couple dozen enhancers highly conserved between human and shark, and pretty much all animals in between.

020

Boris Lenhard @borislenhard.bsky.social · 04/10/2026

Search for laptops with Cherry MX switch keyboards. There are multiple options.

010

Boris Lenhard @borislenhard.bsky.social · 15/09/2026

"Geneticists, phycisists or engineers don't self-regulate." This is incorrect and historically ignorant. See e.g. en.wikipedia.org/wiki/Asiloma... And philosophers, economists and lawyers without adequate knowledge of the domain they wish to regulate can only do more harm than benefit.

en.wikipedia.org

Asilomar Conference on Recombinant DNA - Wikipedia

000

Boris Lenhard @borislenhard.bsky.social · 15/09/2026

... You misrepresented my own stances multiple times already. So please do not give me "category error" - you keep jumping between categories without even noticing. I am not against AI regulation - not at all. I am against stupid AI regulation. And that is where we have talked past each other.

100

Boris Lenhard @borislenhard.bsky.social · 15/09/2026

... You will prevent no autism while failing to prevent preventable diseases at the same time. Based on the exchange here, you sound like you are uninterested in expert consensus or any efforts to reach it that are ongoing. You are uninterested in ulterior motives of people who you agree with. ...

100

Boris Lenhard @borislenhard.bsky.social · 15/09/2026

...and to do it, he has surrounded himself with quacks and sycophants who do not contradict him. There is no category error. Trying to regulate AI based on strongly held wrong beliefs about it is no different than trying to "regulate" vaccines to prevent them from causing autism. ...

100

Boris Lenhard @borislenhard.bsky.social · 15/09/2026

You really have a knack for missing the point. I gave RFK Jr not because of how he plans to use AI, but because he wants to regulate an area he is utterly clueless about, has strong and utterly wrong opinions about, and creates or scraps regulation based on those wrong opinions...

100

Boris Lenhard @borislenhard.bsky.social · 14/09/2026

And, yes, when it comes to risk assessment and what’s true, expertise should trump a majority opinion of non-experts formed on social networks. Sure, the latter can elect a madman to implement their misinformed fantasies instead. We are watching how it is working in real time.

000

Boris Lenhard @borislenhard.bsky.social · 14/09/2026

If you don’t understand what you insist on overseeing, and you nevertheless have strong opinions on it, you will make a dog’s dinner out of it . See what RFK Jr is doing to the US public health now.

110

Boris Lenhard @borislenhard.bsky.social · 14/09/2026

Where did I say we should continue with the current lack of any regulatory framework? I was just saying that we should not agree to a framework that either gives the current market leaders a future advantage, or to one that is swayed by paranoid scenarios into harming the beneficial applications.

000

Boris Lenhard @borislenhard.bsky.social · 14/09/2026

..And when there is no consensus, you should NOT force the society to go with the most paranoid scenario - you don't do that with pharma, vaccines, self-driving cars, or the Large Hadron Collider - and you shouldn't do it with LLMs, either. Doing so is guaranteed to produce more harm than necessary.

010

Boris Lenhard @borislenhard.bsky.social · 14/09/2026

... And until the consensus of experts, the view of non-experts is purely an aesthetic preference. They will choose what sounds better to their worldview, which is completely disconnected from whether it is true or not...

210

Boris Lenhard @borislenhard.bsky.social · 14/09/2026

You misunderstood me then. I said that non-experts do not get to choose their experts. I am not an AI computer scientist. I am a biologist (who uses AI for work) and can judge the validity of suggested biological threats from AI (in short: overwhelmingly nonsense from people who know no biology)...

120

Boris Lenhard @borislenhard.bsky.social · 14/09/2026

...and wants to prove the expert wrong instead, forget it - whatever the expert says, the non-expert will not have the knowledge to understand but will think they have the right to reject it because it clashes with what they were misinformed into believing is true. They already have "their experts".

100

Boris Lenhard @borislenhard.bsky.social · 14/09/2026

That is what Feynman said, and the past pandemic has taught us that it is just not true. The only way for an expert to impart knowledge to a non-expert is if the non-expert has come to learn and accept the expertise of the former. If the non-expert has already decided what they think is true...

100

Boris Lenhard @borislenhard.bsky.social · 14/09/2026

LLMs in the wrong hands can cause a lot of cybersecurity issues, because they can only access (networked) computers. They cannot mobilise physical infrastructure in a dystopian way because the functionality is physically not there, nor is the functionality to build that functionality autonomously.

100

Boris Lenhard @borislenhard.bsky.social · 14/09/2026

The AI people throwing around probabilities of AI causing human extinction involve a chain of guesses, some of which they are clearly not experts in. They are not experts in risk analysis, biology, or on how LLMs can interact with the physical world and how they can not.

100

Boris Lenhard @borislenhard.bsky.social · 14/09/2026

My expertise just allows me to understand some things that people without it do not. And while I can try to explain some of it to non-experts, it is impossible to make up for what takes years in five minutes. Especially to non-experts who do not want to learn, but want to prove me wrong instead.

210

Boris Lenhard @borislenhard.bsky.social · 14/09/2026

P.S. To be able to question the experts, you need to be able to understand the answers they give you. Lack of domain expertise often prevents this. A colleague shared this paper (one author is a legal expert): papers.ssrn.com/sol3/papers.... . The screenshot summarises the point I have made here.

000

Boris Lenhard @borislenhard.bsky.social · 14/09/2026

...they are basically pulled out of thin air. There is no science or expertise whatsoever involved in making those estimates. But they do make for highly emotional newspaper headlines and social media posts. And, before accusing others of rudeness and ad hominem, please examine your own reactions.

000

Boris Lenhard @borislenhard.bsky.social · 14/09/2026

...and you are ignoring them completely. At the same time, "doomers" themselves admit that the current technology is not capable of creating existential risk but just fear that something will "switch" by itself. The percentage probabilities of wiping humans are based on no calculation at all, ...

110

Boris Lenhard @borislenhard.bsky.social · 14/09/2026

I am sorry, but you ARE a screamer. You want to ban or slow down something that you don't understand at all. There are top experts on AI who think that the existential risk for humanity is near-nonsense (LeCun, Marcus, Ng...), who understand the matter division-by=zero times better than you...

210

Boris Lenhard @borislenhard.bsky.social · 14/09/2026

Basically when it suits him to slow down, he will mobilise screamers like you to prophesise doom otherwise. When it will suit him to accelerate, he will accuse the “careful” that they are blocking the next cure for cancer. I advise you to try not to be a useful fool for either narrative.

110

Boris Lenhard @borislenhard.bsky.social · 14/09/2026

I think you have missed my point completely. When you are cautious about something that may not exist, you might end up blocking a lot of good that a technology makes possible. The same Dario Amodei who wants to slow down declared that AI will cure most diseases in a decade. Which is it?

100

Boris Lenhard @borislenhard.bsky.social · 14/09/2026

At the current stage, you are afraid of zombies. Others may scream that zombies are coming, but they are just as clueless on how they might materialise as you are. Any concrete anti-zombie measures are supposed to protect from something that does not exist and that nobody knows how it will look.

000

Boris Lenhard @borislenhard.bsky.social · 14/09/2026

When experts disagree on the level and type of risk and there is no objective way to assess the far-fetched scenarios, it is not up to non-experts to take sides, or choose their “own” experts, either. And yet here you are. You don’t know what you are asking for, really, nor does anyone else.

200

Boris Lenhard @borislenhard.bsky.social · 13/09/2026

No. But we know how to test medications. Describe what an AI equivalent of a three-stage, double-blind clinical trial would be, and I might believe that it is OK to establish the oligopoly where companies that emerged as winners in unregulated conditions get to keep that advantage in perpetuity.

110

Boris Lenhard @borislenhard.bsky.social · 13/09/2026

He should have stuck to funny gags then.

000

Boris Lenhard @borislenhard.bsky.social · 13/09/2026

The precautionary principle often leads to braindead catch-22: slowdowns and moratoria “until we learn more” prevents the collection of the data from which we could possibly learn more. Regulatory capture might lead to AI industry turning into big pharma - the big few controlling the blockbusters.

110

Boris Lenhard @borislenhard.bsky.social · 13/09/2026

OK. Let’s try it and see what happens. Some people are destined to learn from their own mistakes only.

110

Boris Lenhard @borislenhard.bsky.social · 13/09/2026

I agree, and I agree it should be regulated, too. Just not by emotions. Slow it down by expensive regulation, and you are locking into the current oligopoly. That will make Amodei, Altman and Musk much happier than you think, and much happier than they should be.

110

Boris Lenhard @borislenhard.bsky.social · 13/09/2026

Toddlers are wielding it, and toddlers are shouting about how it should be stopped, too. Both have potential to do damage, and latter is more certain at the moment.

000

Boris Lenhard @borislenhard.bsky.social · 13/09/2026

Please don’t shout at me. I am neither an AI seller nor a regulator. I do use it daily in my work because I cannot afford not to. As such, I have some experience with risk assessment. I am stating my opinion, and what you do with it is up to you.

000

Boris Lenhard @borislenhard.bsky.social · 12/09/2026

Yes, it is normal to run an experiment the outcome of which you don’t know. That is what experiments are for. Nothing done at the moment can wipe out the human species or self-evolve to do it. I am all for caution, but non-experts raised on dystopian SF are hopeless at risk assessment.

110

Boris Lenhard @borislenhard.bsky.social · 12/09/2026

A full moratorium will not happen. It would block too many genuinely beneficial developments. In science applications, you get to “supervise” and use many not fully understood phenomena. Unknown unknowns are everywhere, and we learn about them as we go along.

100

Reposted by Boris Lenhard

Adam Bienkov @adambienkov.bsky.social · 07/09/2026

Watching BBC News and it's striking to hear the presenters repeatedly referring to the AfD as "far-right" whilst refusing to ever say the same about the hundreds of actual masked blackshirts currently swarming British ports

8743341048

Reposted by Boris Lenhard

Thangam Debbonaire @thangamdebbonaire.bsky.social · 08/09/2026

A very useful guide from @stephenkb.bsky.social to the difference between legitimate concerns and racism, and how that muddled approach leads to muddled policy. Also. Racist thuggery isn’t protest. giftarticle.ft.com/giftarticle/...

giftarticle.ft.com

Labour’s muddled migration policy won’t stop thuggery

A government that can’t distinguish between ‘legitimate concerns’ and open racism is only making things worse

28626

Boris Lenhard @borislenhard.bsky.social · 28/08/2026

I honestly don't understand how people who choose to work for the Telegraph, the Mail or the Express justify to themselves the morality of contributing to their vileness. Surely they cannot all be natural-born and -bred moral degenerates? Or can they?

010

Boris Lenhard @borislenhard.bsky.social · 28/08/2026

The people they are posturing for wouldn't notice this if their lives depended on it.

000

Reposted by Boris Lenhard

Franglaise @franglaise.bsky.social · 27/08/2026

Richard Tice is going to pop up any minute to tell the people of Nepal they should just enjoy the swim

432251

Reposted by Boris Lenhard

Adam Rutherford @adamrutherford.bsky.social · 26/08/2026

Pahahahahah Cofnas

33687127

Boris Lenhard @borislenhard.bsky.social · 24/08/2026

Because no sane university in good standing would ever hire him to do that.

131

Boris Lenhard @borislenhard.bsky.social · 24/08/2026

Cofnas has no new, methodologically sound psychometric data, and as a philosopher, he does not engage in any activity that would produce it. He is just defending old arguments with some new arguments of abysmal quality, without addressing the problems that got the former rejected in the first place.

190

Boris Lenhard @borislenhard.bsky.social · 24/08/2026

His "research" in the field of 'race realism' is not research. 'Race realism' is not a field but a collection of ideas based on confounded studies, discarded by the overwhelming majority of psychologists and psychometric researchers. Whoever wanted to revive it would need to show some new data.

160

Boris Lenhard @borislenhard.bsky.social · 24/08/2026

This. If you are trying to rehabilitate a hypothesis that the overwhelming consensus has thoroughly buried, the minimum requirement is to have new data to support that rehabilitation attempt. Writing essays and blogs based on the reasoning of debunked racist cranks will just make you one of them.

010

Boris Lenhard @borislenhard.bsky.social · 21/08/2026

This sounds less like Henry VIII and more like Chairman Mao.

010

Boris Lenhard @borislenhard.bsky.social · 21/08/2026

Wasn't rhetoric originally part of the Trivium, along with logic (basically a precursor to critical thinking) and grammar?

030

Boris Lenhard @borislenhard.bsky.social · 21/08/2026

This doesn’t require much effort any more - Claude Code or OpenAI Codex can download a page from the original site, insert a fake story in it, and deploy it, with only a few prompts.

030