HNHacker News
TopNewBestAskShowJobs

elliotto

735 karma · joined September 15, 2020

Software engineer interested in machine learning and data science.
submissionscomments
elliotto··on A non-anthropomorphized view of LLMs
My personal views are an animist / panpsychist / pancomputationalist combination drawing most of my inspiration from the works of Joscha Bach and Stephen Wolfram (https://writings.stephenwolfram.com/2021/03/what-is-consciou...). I think that the underlying substrate of the universe is consciousness, and human and animal and computer minds result in structures that are able to present and tell narratives about themselves, isolating themselves from the other (avidya in Buddhism). I certainly don't claim to be correct, but I present a model that others can interrogate and look for holes in.

Under my model, these systems you have described are conscious, but not in a way that they can communicate or experience time or memory the way human beings do.

My general list of questions for those presenting a model of consciousness are: 1) Are you conscious? (hopefully you say yes or our friend Descartes would like a word with you!) 2) Am I conscious? How do you know? 3) Is a dog conscious? 4) Is a worm conscious? 5) Is a bacterium conscious? 6) Is a human embryo / baby consious? And if so, was there a point that it was not conscious, and what does it mean for that switch to occur?

What is your view of consciousness?

elliotto··on A non-anthropomorphized view of LLMs
To claim that LLMs do not experience consciousness requires a model of how consciousness works. The author has not presented a model, and instead relied on emotive language leaning on the absurdity of the claim. I would say that any model one presents of consciousness often comes off as just as absurd as the claim that LLMs experience it. It's a great exercise to sit down and write out your own perspective on how consciousness works, to feel out where the holes are.

The author also claims that a function (R^n)^c -> (R^n)^c is dramatically different to the human experience of consciousness. Yet the author's text I am reading, and any information they can communicate to me, exists entirely in (R^n)^c.

elliotto··on LLMs should not replace therapists
I have had a similar interaction when I was building an AI agent with tool use. It kept on telling me it was calling the tools, and I went through my code to debug why the output wasn't showing up, and it turns out it was lying and 'hallucinating' the response. But it doesn't feel like 'hallucinating', it feels more like fooling me with responses.

It is a really confronting thing to be tricked by a bot. I am an ML engineer with a master's in machine learning, experience at a research group in gen-ai (pre-chatgpt), and I understand how these systems work from the underlying mathematics all the way through to the text being displayed on the screen. But I spent 30 minutes debugging my system because the bot had built up my trust and then lied to me that it was doing what it said it was doing, and been convincing enough in its hallucination for me to believe it.

I cannot imagine how dangerous this skill could be when deployed against someone who doesn't know how the sausage is made. Think validating conspiracy theories and convincing humans into action.

elliotto··on I extracted the safety filters from Apple Intelligence models
Unalive and other self censors were adopted by young people because the tiktok algorithm would reprioritize videos that included specific words. Then it made its way into the culture. It has nothing to do with being performative
elliotto··on Writing documentation for AI: best practices
An effect I've found of building agents that interacts with an API is that the bot serves as a smoke test for error handling. If your API doesn't return a clear and actionable error, or doesn't validate json fields clearly or correctly, the bot will end up finding it out and getting confused.

I can imagine a near future where crud endpoints are just entirely tested by an AI service that tries to read the docs and navigate the endpoints and picks up any inconsistencies and faults.

elliotto··on When Fine-Tuning Makes Sense: A Developer's Guide
A really simple blog post for any task that you think is worthwhile would be enough to move the field forward. The blog post should include:

1) the training configuration and code 2) the data used to fine tune 3) a set of input/output comparisons comparing the tuned bot to the original bot that show it's learned something interesting

For something really compelling it would host the created models on a repo that I could download and use. The gold standard would be to host them and provide a browser interface, but this could be expensive for gpu costs.

This blog post currently doesn't exist, or if it does I haven't been able to find it in the sea of medium articles detailing an outdated hugging face api

elliotto··on When Fine-Tuning Makes Sense: A Developer's Guide
Chiming in here to say that I was tasked to implement a fine tuning method for my AI startup and I also couldn't find any actual implemented outputs. There are piles of tutorials and blog posts and extensive documentation on hugging face transformers about the tools provided to do this, but I was unable to find a single demonstration of 'here is the base model output' vs 'here is the fine tuned output'. Doesn't have to be online like you suggested, even a screenshot or text blob showing how the fine tuning affected it would be useful.

I am in a similar boat to you where I have developed a great sense for how the bots will respond to prompting and how much detail and context is required because I've been able to iterate and experiment with this. But have no mental model at all about how fine tuning is meant to perform.

elliotto··on OpenAI can stop pretending
I would encourage you to read books like Varoufakis' Technofeudalism and Piketty's Capital in the 21st Century. The distribution of wealth and capture of profits has changed significantly and no longer resembles the economic environment of previous generations.
elliotto··on OpenAI can stop pretending
This is true under an economic system that appropriately distributes the fruits of economic production to workers; perhaps one that existed in the post-ww2 to neoliberal era. But it is untrue under a rentier / technofeudal / oligarch economy, in which all excess profits are captured by a cabal of the owner class. Which of these states describes the modern economy is a political exercise.
elliotto··on OpenAI can stop pretending
The fact that productivity gains are bad for employees (labour vs capital) should raise significant questions about the economic systems that we have in place.

Instead, AI is treated as the boogeyman and progress is treated as the devil. These events should be an impetus towards class consciousness, but instead the hate is directed to the token producing machine. I wonder how much of this redirection is a deliberate psyop.

elliotto··on xAI's Grok 3 comes to Microsoft Azure
Didn't this guy hit the salute in front of the entire world? To me it seems very likely that he would inject a racist prompt. Far more likely than a random hacker doing so to discredit him.
elliotto··on xAI's Grok 3 comes to Microsoft Azure
It's difficult to make the claim that the AI not complying with a racist prompt is a positive signal for the organisation that wrote the racist prompt.
elliotto··on Game theory illustrated by an animated cartoon game
The success of tit-for-tat is a common misunderstanding; it is only the most successful within the makeup of Axelrod's tournament. The OP explains that the strategy is entirely dependent on the environment, and tit-for-tat may not always be the optimal strategy.
elliotto··on Burrito Now, Pay Later
Terrible post, seemingly chatgpt generated for large sections (see chapter What Happens When Borrowers Miss Payments) and disappointing to see on the front page of HN. This is the quality and calibre of analysis I would expect to find on a LinkedIn feed post.
elliotto··on DoorDash to acquire Deliveroo
I have a high moral high ground because I don't use these services at all due to poor treatment of employees.

It's difficult to conflate the moral high ground with maximal participation in a weird auction format when people are just trying to buy a burrito.

In most other countries (like mine Australia) the price of the good is the advertised price. There is no extra weird moral based financial game played on top.

elliotto··on Librarians are dangerous
A writing style like this indicates that the author does not have the taste to write well. This is a signal that the content will not be good.
elliotto··on Cardiac arrest deaths among marathoners have decreased, study finds
Can't read the actual study because it's behind a medical journal paywall. It looks like they speculate that improved CPR training and AED use has caused this.

My speculation would be that these mega running nerds all wear watches that track their heart rate 24/7 now. I suspect if something is wrong then you get an earlier detection rather than only discovering it on a race day.

elliotto··on Layoffs Don't Work
When I hear things like this, where one person has been given command of the plane and tanked it straight into the ground, with dire consequences particularly for the middle east employees you have mentioned, I am filled with concern about the stability of our systems in general. It often feels like in the history books when you read about the land the King built and then his son inherits it and burns it to the ground. This can't be a sustainable way to run a society.
elliotto··on Layoffs Don't Work
Because societies in which not everyone gets a vote end up collapsing as those with the vote are eaten by those without
elliotto··on The Demoralization is just Beginning
The OP was talking about Shi Zhengrong https://en.wikipedia.org/wiki/Shi_Zhengrong who came to Australia (my country) in the 90s, learned a bunch about photovoltaics, was unable to make a business here due to our failed government policies, then returned to China and made a pv megabusiness.

The truth of the matter is that most cutting edge research is spearheaded by the west. The problem is that the west can't build or prioritise anything useful with it.

elliotto··on The housing theory of everything (2021)
I think you present an idealized version of market economies that is not in line with my experience (I live in Sydney Australia our housing system has collapsed). Our disagreement seems predicated on our beliefs around r > g. Have you read Piketty?
elliotto··on The housing theory of everything (2021)
Your growth rate of 560k to 627k is 11.9% (r - the rate of return on capital). This outpaces wage growth - let's assume generous non-inflation corrected 5 % (g - a proxy for the growth rate of the economy)

Your assets after 50 years: 267x Your wages after 50 years: 11.4x Your new ratio of assets to wages: 23.4

This calc works with any number where r > g (source: capital in the 21st century by piketty). The end result of asset growth outpacing wages is complete divergence and the end game of Monopoly where 1 guy owns everything. Usually by then the system has cracked and the society has burned down.

elliotto··on GPT-4.5
This website sucks but successfully loaded from Aus rn on my phone. It's full of ads - possibly your ad blocker is killing it?
elliotto··on DOGE's only public ledger is riddled with mistakes
I'm from Australia lol the US constantly interferes in our country directly and indirectly. 'Busybody that meddles' is a pretty soft term for it, I would call it aggressively destablises its allies.

If you take away the premise that the US is doing something good, like for PEPFAR, then what is left? A gun wielding maniacal imperial power that you have no reason to deal or treat with. This is bad for the US and bad for the world

elliotto··on DOGE's only public ledger is riddled with mistakes
https://youtu.be/O4xHdUOI7ag?si=RPBNdpAiYHuYe7rf

I would strongly recommend listening to Rory Stewart's commentary in the first half of this podcast about the function and role of USAID as an expression of soft power.

elliotto··on Barcelona buys apartment building at center of eviction protests
'Your' property exists as a line item on a ledger legitimized by violence of the state. In exchange for receiving electricity and water and transport and police support for securing it, you have to follow a few rules.
elliotto··on Migraine is more than a headache – a rethink offers hope
https://archive.md/Apbxw
elliotto··on Sharing a ChatGPT Account with My Wife
I think almost all N are a suboptimal solution to what are inherently structural social problems that arise from a distribution of wealth and organization of labor where N is any technological invention of the past 20 years. Open source insulin lol the empire is collapsing.
elliotto··on Sharing a ChatGPT Account with My Wife
Isn't this what Google does already? And it throws ads at you (ranked on whichever company pays them the most)

I suspect a decent open source llm would have a lot more transparency than the current methods of coagulating collective information

elliotto··on Google drops pledge not to use AI for weapons or surveillance
I think the poster is applying your statement about leaving money on the table. Structural requirements to not leave money on the table is a Moloch results that leads to the deterioration of the system into being just stealing as much shit as possible.
← PreviousPage 4 of 6Next →