iDon

I'd agree with other posters - I'm seeing less hallucination; probably the model-makers are tuning to reduce that, because people make a big deal about it.

I see it as a small deal - it reminds us to check the responses against references. LLMs are a statistical construction, and everyone accepts without complaint that statistical models have a predicted false-positive and false-negative rate. I think of "hallucination" as a false positive; false negative is no answer when the model could have made a useful response. I think there's some relation between our creativity and the LLM capability we dismissively call hallucination.

I just went down a rabbit-hole, and happily I found the rabbit - an early example of this type of prompt (draw an unusual animal character in a vector graphics language). There is a paper and a 1 hour podcast resulting from a Microsoft evaluation of a pre-release of GPT4. One of the prompts (see pages 4,7,8 in the paper PDF) was to draw a unicorn in TikZ. (I'll leave to the historians the questions of whether this was the first, or whether Simon Willison may have been inspired by this). I remember hearing some of the podcast, and recalled the prompt about balancing on a nail, and the triangle forming the unicorn's tusk; this was clearly a big step beyond choosing the next word and blending images.

https://arxiv.org/abs/2303.12712

yes, that looks like the one. I think I was using the conditional jump, so it could have been djnz, but I think it was jr. 1 byte jump offset, and I recall my little function bumping against that limit - I might have re-arranged it. http://z80-heaven.wikidot.com/instructions-set:jr

The Z80 has a relative jump. Teenage me was loading a hand-assembled search function as an initialised array of integers in Basic, and relative jumps didn't need to be adjusted for the load location, which reduced the amount of post-load adjusting of addresses. It was +100x faster than Basic, just a straight scan.

A language like E seems about the right level for a simple user-friendly programmatic layer wrapping applications, i.e. enable the users to ask a natural language question or build a simple GUI layer which has question actions, with results displayed in the custom GUI or exported as files (CSV or spreadsheets etc).

I expect we'll see a blurring of the web app GUI, allowing users and LLMs to augment the GUI.

There are a few tools offering the GUI part of that; I had read about Google A2UI, and search finds more : Anthropic Claude Rivet, CopilotKit (uses AG-UI Standard), Chainlit (Python UI framework), Vercel AI SDK (React Server Components stream from the LLM to the client). URLs : https://github.com/google/A2UI, https://fast.io/resources/best-ui-frameworks-ai-agents/, https://rivet.ironcladapp.com/ and https://techcommunity.microsoft.com/blog/azure-ai-foundry-bl..., https://www.copilotkit.ai/generative-ui, https://www.thesys.dev/blogs/openui, ...

Representing that as a "climate spiral" would make it unnecessary to adjust for the seasons, and the original data could be used instead of a statistical view. It makes it easy for anyone to see the trend. - https://en.wikipedia.org/wiki/Climate_spiral - https://www.theclimatebrink.com/p/visualizing-daily-global-t... - https://www.climate-lab-book.ac.uk/spirals/

He retained the original name, from 2006 ...

https://au.pinterest.com/pin/triple-dragon--1742330793037990... fractaldomains.com Triple Dragon Triple dragon fractal. Equation: f(z) = z^3/(z^3 + 1) + c, c = 0.18 + 0.68 i

https://www.fractaldomains.com/2011/08/triple-dragon/

https://web.archive.org/web/20200000000000*/https://www.frac... https://web.archive.org/web/20200209172651/http://www.fracta...

Fractal Domains Exploring fractals with your Macintosh

Triple Dragon September 3, 2006 Click image to see full size

Julia set using square orbit trap. The formula used was f(z) = z^3/(z^3 + 1) + c, c = 0.18 + 0.68 i Downloads and extras “Triple Dragon” Parameter File

Copyright © 2011 Dennis C. De Mars

triple-dragon.zip : 1058 3-Sep-2006 20:06:36 Triple Dragon DMrsFra2...

---

Paul's image is a nice rendering of an attractive fractal.

> ... It starts by stepping into a shallow pool of golden light. > ... Together they act as both a choir and an audience

I think I'm not the target audience. I guess they are going to need to sign up a lot of people, to train on their scans + their medical outcomes. So the article is talking to people who will get enthused by it, which is more difficult after the question of 23AndMe data sale.

The risk is humans using AI to control / exploit / coerce / injure other humans. The risk of AIs being given enough agency to threaten humans comes after that - they'll only have the agency we give them (being "alive" or "conscious" is not the near-term risk).

The article lists diplomatic actions which might help to manage the risk, starting with "An agreement between America and China" - they all sound like impossible dreams. We had ~80 years of relative peace and prosperity in which to construct a framework of unity to face challenges like AI (and global warming, which until GPT I thought was the bigger risk); international unity is weaker than ever. In geopolitics and defence, capability of other nations is the concern rather than intention; the capability curve of LLMs is heading off our charts. We're backed into tight corners on nuclear proliferation, global warming, and I can see LLM-enabled conflicts (cyber warfare, infrastructure terrorism) pushing us over those other edges. Our democracies seem weakened, and I expect LLMs will empower those using social media to create conflict and control opinion. We're familiar with the cycle of inventing new technology which benefits people, then seeing how long before people invent ways to misuse it. There is a possibility here that LLMs could be used to solve the problems we are juggling, but I struggle to imagine that people won't misuse it even faster.

The article is a start on thinking and talking about managing our risks. The best outcome would be it is so well managed that, like the Y2K "bug", people say "after all that hoopla, nothing happened". I'm not seeing a smooth path to there.

Related (posted just 2hours before this article) : https://news.ycombinator.com/item?id=46086833 "Blimps lifting quantum data centers to the stratosphere? (newatlas.com)" "... blimps, to lift quantum computers to the stratosphere. There, at an altitude of about 20 km (12.4 miles), temperatures are in the -50 °C range (about -58 °F) and would be cold enough to allow the qubits to function correctly."

I have drafted a photo gallery app in Rust / Tauri, using a JavaScript framework in the frontend. The backend can read directories and files directly, and because the backend and frontend are in a single process, the backend simply passes a file handle (path string possibly) to the frontend. In contrast Electron has to send the image file between processes. I started with Electron and I think that was the point I shifted to Rust / Tauri; seeing the images display immediately was a revelation. Rust / Tauri has the advantages of a desktop app, and I have the option to use the frontend as a web app also.

This Python binding (pytauri) is interesting too - I have colleagues with Python functionality they want to surface on the web, and this would give the possibility of running as a desktop app also - good for large datasets.

A search for : "character studies of old farmers" yielded good results. The results are drawings / engravings, which may reflect the balance of the collection, and perhaps this subject is more used in practice than in marketable oil paintings.

Since this is a semantic search, using a vector embedding, it will handle meanings better than a text search, which would handle names better.

The article asks what is the meaning of OpenAI's statement that "the UAE will become the first country in the world to enable ChatGPT nationwide."

My first guess would be that it will be a geo-fenced service, in particular UAE residents will have (subsidised) access to it and perhaps not to the global service, and it will have a system prompt designed and tuned in consultation with the UAE government.

IBM did a partial implementation of ASN.1 in Java, and released it via the IBM AlphaWorks open-source repository. I used it in a telecommunications system in the 90s. Luckily the GSM protocol we were interfacing with only used a small subset of ASN.1, which was covered by the IBM software. IBM AlphaWorks is still online : https://www.ibm.com/support/pages/aix-toolbox-open-source-so... but only lists libtasn1, which is in C.

Here's a post about AlphaWorks : https://www.cnet.com/tech/services-and-software/ibm-alphawor...

Searching for that, I found this post, https://stackoverflow.com/questions/37056554/opensource-java... which mentions : https://www.beanit.com/asn1/ https://sourceforge.net/projects/jac-asn1/ which are more recent java ASN.1 implementations.

The name I recall from using it on a PDP 11/04 was 'Text Editor and Corrector' https://en.wikipedia.org/wiki/TECO_(text_editor)

I vaguely recall it had a line-open / visual mode, like ex/vi, which we didn't use because we were on a dot-matrix line printer / teletype. The ADM-3A had the Ctrl key on the home line; this design made it easy for editors from that period (vi, emacs) to make heavy use of Ctrl.

Thanks to those who posted bits of TECO - I'd forgotten how the character movement was similar to vi. A fellow student in our CS honours year had a semester thesis project analysing the grammar of vi commands and specifying it in a formal grammar. The combination of action x movement is powerful, simple and concise.

For decades people have been dreaming of higher-level languages, where a user can simply specify what they want and not how to do it (the name of the programming language Forth derives from '4th Generation Language', reflecting this idea).

Here we are - we've arrived at the next level.

The emphasis in my prompts is specification : clear and concise, defining terms as they are introduced, and I've had good results with that. I expect that we'll see specification/prompt languages evolve, in the same way that MCP has become a defacto standard API for connecting LLMs to other applications and servers. We could use a lot of the ideas from existing specification languages, and there has been a lot of work done on this over 40+ years, but my impression is they are largely fairly strict, because their motivation was provably-correct code. The ideas can be used in a more relaxed way, because prompting fits well with rapid application development (RAD) and prototyping - I think there is a sweet spot of high productivity in a kind of REPL (read/evaluate/print loop) with symbolic references and structure embedded in free-form text.

Other comments have mentioned the importance of specification and requirements analysis, and dahlfox menions being able to patch new elements into the structure in subsequent prompts (via BASIC line number insertion).

I am running this OK in Ubuntu 2404 : https://github.com/aaddrick/claude-desktop-debian Claude Desktop for Debian-based Linux distributions

From Claude I have connected to these MCP servers OK : @modelcontextprotocol/server-filesystem, @executeautomation/playwright-mcp-server.

I have connected to OP's extension (browsermcp.io) from vsCode (and clicked 1 tab button OK), but not from Claude desktop so far (I get Cannot find module 'node:path'; which is require-d in npm/lib/cli.js; tried node 18,20,22; some suggestions here : https://medium.com/@aleksej.gudkov/error-cannot-find-module-... ).

I use this for quick comparison of minor differences in code / data (e.g. tables, DNA sequences). Variations in indentation are highlighted also, by standing out of the plane.

It's good to hear reports of successful viewing. I've got a 3d / stereogram photo gallery app on the back burner; sounds like a reasonable number of people would be able to view it. There are plenty of guides on how to learn this; some are linked here https://www.reddit.com/r/CrossView/wiki/index/. You-tube used to have support for this; there are still videos tagged yt3d - just regular videos now, not interlaced.

Thanks to OP and other posters - various ideas useful in different cases.

The xargs idea made me think of using bash as the parser :

  bash -c "exec -c bash -c 'source $CONFIG/main.bash; env'"

This test .bash file contains multiple source-s of other .bash files, which contain a mix of comments, functions, set and env vars - just the env vars are exported by env. This seems useful e.g. for collating & summarising an environment for docker run -e.

This outputs the env vars to stdout; for the OP's purpose, the output could be sourced :

  envFile=$(mktemp /tmp/env.XXXXXX);

  bash -c "exec -c bash -c 'source $CONFIG/main.bash; env'" > $envFile;

  env $(cat $envFile) sh -c 'echo $API_HOST'

# For Bourne shell, use env -i in place of exec -c :

sh -c "env -i sh -c '. $CONFIG/main.sh; env'" > $envFile

Extending that idea to the web, or at least to the blogosphere and information / knowledge web-sites, seems useful. I wonder if there is a web service which has calculated vector embeddings for some of the web, and supports vector search, e.g. given a URL, find URLs with similar embeddings. Inverting that, web-sites could annotate their web pages with embeddings via json-ld; which search engines could utilise. Both these ideas might be impractical, e.g. the cost of http GET of the vector might be similar to the cost of calculating the embedding; and the embedding would be only comparable with embeddings from the same model (which would be recorded in the json-ld) so it would age quickly. It would also be subject to SEO gaming, like meta tags.

A quick search didn't find either of these; the closest was this paper which used json-ld to record a vector reduced to 2 dimensions using tSNE : https://hajirajabeen.github.io/publications/Metadata_for_Eme... Metadata standards for the FAIR sharing of vector embeddings in Biomedicine S¸ enay Kafkas et al.

D3 blocks can be viewed by replacing gist.github.com or bl.ocks.org with blocks.roadtolarissa.com, e.g. for your link : https://blocks.roadtolarissa.com/AndreaSimeone/1e9ed38b46b95... (the home page https://roadtolarissa.com/ has a collection of visualisations, with an RSS link for updates, and the author posts some here as 1wheel).

The simulation playback in your link "Force Graph Trails" would be useful when trying to refine and balance the forces in a force graph; the small amount I've done felt like juggling parameters, searching in a space with too many dimensions.

I used the Cheq app to make UPI payments via smartphone / QR code, during a holiday in India in July. The app listed locations for completing the Know Your Customer (KYC) process - I did this at Transcorp in Bengaluru. It was worth the effort - it is far more convenient than cash in most cases. Most vendors kept little change, and even restaurants were not keen on receiving a 500R note for a 200R meal. Similarly for museums, art galleries, historic buildings. Auto-rickshaw drivers were happy with cash, and had a reasonable amount of change. I did find a couple of vendors who weren't registered as merchants for UPI, and I could only transfer to merchants, not other users. For travellers I'd suggest setting up UPI beforehand if possible; as mentioned you can do it as you fly in (Thomas Cook offers this service in the airport), but the KYC might take a while.

(Side-tracking from payments to SIMs : ) Several historic buildings had an online-only ticket process with a QR code displayed outside. The user experience of the websites was variable (navigating the forms and input widgets on a phone while standing in the street could be trying), and they required a lot of information, eg. Passport number and sometimes home address, usually including authentication via one time password sent to mobile phone number and possibly they only work for Indian SIMs. Those sites accepted credit card payment, including international. I couldn't count the number of times I gave my phone number, for all sorts of purposes, e.g. when checking in my phone at a temple, I was photographed and gave my number - simple id and method for finding the phone among all the others (when I realised that I switched the phone back on). And for all of these a local phone number is required, so pick up a local SIM at the airport or in the city. I did both, and I think the airport process was quicker. They do require full id for SIMs, including a local address, e.g. a hotel. A 28-day SIM cost only 299 rupees so I got a second one for increased coverage and capacity, because the networks are over-subscribed. (That's less than the daily rate at home). A dual-SIM phone makes that easy, and I could also run with one local SIM and my home SIM to receive SMSs. I was able to download maps and apps while travelling on buses and trains, and use tracking apps - which helped with knowing when a bus was approaching my stop.

Agriculture Victoria (DEECA) : Big data in agriculture | Web Administrator/Developer | Data Engineer | Melbourne (Australia) | $91,418 to $103,725 + super. | Full-time

Innovative web applications : development and deployment of cutting-edge open-source bioinformatics tools for the integration and visualisation of large multi-dimensional genomic and phenotypic datasets. These tools are used by researchers to understand the genetic architecture and molecular mechanisms underlying trait variation to support breeding of crops important to the Victorian and Australian agricultural sectors.

---

## Web Administrator/Developer :

Required : - a bachelor's degree in computer-science, mathematics, bioinformatics, or other suitably related field - broad skills in server administration and modern back- and front-end JavaScript frameworks - Linux / UNIX server administration and maintenance skills, including in cloud environments; familiarity with JavaScript frontend, backend and visualisation frameworks such as Ember.js and D3.js; advanced scripting and experience with code version control systems - backend / API development technologies with a preference for Node.js frameworks.

Details and application : https://www.seek.com.au/job/66285739?type=standout#sol=5508f...

---

## Data Engineer : Required : - JavaScript, Node.js, MongoDB or other database experience, - A degree in computational biology, computer science, mathematics or a related discipline.

Relevant skills and experience : - backend / API development - Express.js / Loopback.js and other Node.js back-end frameworks - Python, R, C++, Java

Details and application : https://www.seek.com.au/job/66286845?type=standout#sol=34231...

---

Hybrid work arrangements (3 days onsite, 2 days WFH) Fixed term contract position available for a period of 3 years.

The applications require responses to the Key Selection Criteria.

Project Repositories : https://github.com/plantinformatics

Agriculture Victoria (DEECA) : Big data in agriculture | Web Administrator/Developer | Data Engineer | Melbourne (Australia) | $91,418 to $103,725 + super. | Full-time

Innovative web applications : development and deployment of cutting-edge open-source bioinformatics tools for the integration and visualisation of large multi-dimensional genomic and phenotypic datasets. These tools are used by researchers to understand the genetic architecture and molecular mechanisms underlying trait variation to support breeding of crops important to the Victorian and Australian agricultural sectors.

---

## Web Administrator/Developer :

Required : - a bachelor's degree in computer-science, mathematics, bioinformatics, or other suitably related field - broad skills in server administration and modern back- and front-end JavaScript frameworks - Linux / UNIX server administration and maintenance skills, including in cloud environments; familiarity with JavaScript frontend, backend and visualisation frameworks such as Ember.js and D3.js; advanced scripting and experience with code version control systems - backend / API development technologies with a preference for Node.js frameworks.

Details and application : https://www.seek.com.au/job/66285739?type=standout#sol=5508f...

---

## Data Engineer : Required : - JavaScript, Node.js, MongoDB or other database experience, - A degree in computational biology, computer science, mathematics or a related discipline.

Relevant skills and experience : - backend / API development - Express.js / Loopback.js and other Node.js back-end frameworks - Python, R, C++, Java

Details and application : https://www.seek.com.au/job/66286845?type=standout#sol=34231...

---

Hybrid work arrangements (3 days onsite, 2 days WFH) Fixed term contract position available for a period of 3 years.

The applications require responses to the Key Selection Criteria.

Project Repositories : https://github.com/plantinformatics

This blog is reporting on doing the advent of code via ChatGPT : https://note89.github.io/the-advent-of-code-ai-edition/

Not OP, but I found various info about the values of Nigella as a herbal remedy, e.g. with garlic https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4687760/

When I read "black seed" in the comment I guessed it might be Nigella seeds, which are sprinkled atop pides in the local middle-eastern bakeries, adding a lovely piquancy to the bread.

From : https://www.sri.com/engage/ventures/siri

https://www.sri.com/work/timeline-innovation/timeline.php?ti...

"Siri, the first virtual personal assistant, arose from decades of SRI research in artificial intelligence (AI). The technology was developed through the SRI-led Cognitive Assistant that Learns and Organizes (CALO) project within DARPA's Personalized Assistant that Learns (PAL) program, the largest-known AI project in U.S. History, and joint work with EPFL, the Swiss institute of technology.

SRI spun off Siri, Inc. in 2007 to bring the technology to consumers, raising $24 million in two rounds of financing.

In April 2010, Apple acquired Siri, and in October 2011, Siri was unveiled as an integrated feature of the Apple iPhone 4S."

I'm not verifying that - I just found it via google search "sri Siri"

I like the link to Logan's Run :-), it could be also true.