From mathematical insights to practical applications: Two perspectives on AI
19th April 2025
As AI continues to transform our technological landscape, two recent books offer distinct yet complementary perspectives on understanding and working with these powerful tools. Stephen Wolfram's technical deep dive and Ethan Mollick's practical guide approach the subject from different angles, but both provide valuable insights for navigating our AI-integrated future.
- What is ChatGPT Doing?: Wolfram's Technical Lens
Stephen Wolfram's exploration of large language models is characteristically thorough and mathematically oriented. While dense in parts, his analysis reveals fascinating insights about both AI and human cognition.
Perhaps most intriguing is Wolfram's observation that generative AI unexpectedly teaches us about human language production. These systems, in modelling our linguistic patterns with such accuracy, hold up a mirror to our own cognitive processes, perhaps revealing structures and patterns we had not fully appreciated before.
Wolfram does not shy away from highlighting limitations, particularly regarding computational capabilities. As sophisticated as next-word prediction has become through multi-billion parameter neural networks, these systems fundamentally lack true mathematical reasoning. However, his proposal of integrating language models with computational tools like WolframAlpha presents an elegant solution, combining the conversational fluency of AI with precise computational power.
- Co-intelligence: Mollick's Practical Framework
Ethan Mollick takes a decidedly more accessible approach in "Co-intelligence," offering accessible strategies for effective human-AI collaboration across various contexts. His framework includes several practical principles:
- Invite AI to the table as a collaborator rather than merely a tool
- Maintain human oversight and decision-making authority
- Communicate with AI systems as if they were people with specific roles
- Assume current AI represents the lowest capability level you will work with going forward
What makes Mollick's work particularly valuable is its contextual applications. Drawing from his background as a business professor, he methodically examines how these principles apply across different collaborative scenarios: from personal assistant to creative partner, coworker, tutor, coach, and beyond. With a technology, that, even now, retains some of the quality of a solution looking for a problem, these grounded suggestions act as a counterpoint to the torrent of hype that that deluges our working lives, especially if you frequent LinkedIn a lot as I am doing at this time while searching for new freelance work.
- Complementary Perspectives
Though differing significantly in their technical depth and intended audience, both books contribute meaningfully to our understanding of AI. Wolfram's mathematical rigour provides theoretical grounding, while Mollick's practical frameworks offer immediate actionable insights. For general readers looking to productively integrate AI into their work and life, Mollick's accessible approach serves as an excellent entry point. Those seeking deeper technical understanding will find Wolfram's analysis challenging but rewarding.
As we navigate this rapidly evolving landscape, perspectives from both technical innovators and practical implementers will be essential in helping us maximise the benefits of AI while mitigating potential drawbacks. As ever, the hype outpaces the practical experiences, leaving us to suffer the marketing output while awaiting real experiences to be shared. It is the latter is more tangible and will allow us to make use of game-changing technical advances.
Removing query strings from any URL on an Nginx-powered website
12th April 2025
My public transport website is produced using Hugo and is hosted on a web server with Nginx. Usually, I use Apache, so this is an exception. When Google highlighted some duplication caused by unneeded query strings, I set to work. However, doing anything with URL's like redirection cannot use a .htaccess file or MOD_REWRITE on Nginx. Thus, such clauses have to go somewhere else and take a different form.
In my case, the configuration file to edit is /etc/nginx/sites-available/default because that was what was enabled. Once I had that open, I needed to find the following block:
location / {
# First attempt to serve request as file, then
# as directory, then fall back to displaying a 404.
try_files $uri $uri/ =404;
}Because I have one section for port 80 and another for port 443, there were two locations that I needed to update due to duplication, though I may have got away without altering the second of these. After adding the redirection clause, the block became:
location / {
# First attempt to serve request as file, then
# as directory, then fall back to displaying a 404.
try_files $uri $uri/ =404;
# Remove query strings only when necessary
if ($args) {
rewrite ^(.*)$ $1? permanent;
}
}The result of the addition is a permanent (301) redirection whenever there are arguments passed in a query string. The $1? portion is the rewritten URL without a query string that was retrieved in the initial ^(.*)$ portion. In other words, the redirect it from the original address to a new one with only the part preceding the question mark.
Handily, Nginx allows you to test your updated configuration using the following command:
sudo nginx -t
That helped me with some debugging. Once all was in order, I needed to reload the service by issuing this command:
sudo systemctl reload nginx
With Apache, there is no need to restart the service after updating the .htaccess file, which adds some convenience. The different locations also mean some care with backups when upgrading the operating system or moving from one server to another. Apart from that, all works well, proving that there can be different ways to complete the same task.
The critical differences between Generative AI, AI Agents, and Agentic Systems
9th April 2025
The distinction between three key artificial intelligence concepts can be explained without technical jargon. Here then are the descriptions:
- Generative AI functions as a responsive assistant that creates content when prompted but lacks initiative, memory or goals. Examples include ChatGPT, Claude and GitHub Copilot.
- AI Agents represent a step forward, actively completing tasks by planning, using tools, interacting with APIs and working through processes independently with minimal supervision, similar to a junior colleague.
- Agentic AI represents the most sophisticated approach, possessing goals and memory while adapting to changing circumstances; it operates as a thinking system rather than a simple chatbot, capable of collaboration, self-improvement and autonomous operation.
This evolution marks a significant shift from building applications to designing autonomous workflows, with various frameworks currently being developed in this rapidly advancing field.
Add Canonical Tags to WordPress without plugins
31st March 2025
Search engines need to know which is which because they cannot know which is the real content when there is any duplication, unless you tell them. That is where canonical tags come in handy. By default, WordPress appears to add these for posts and pages, which makes sense. However, you can add them for other places too. While a plugin can do this for you, adding some code to your theme's functions.php file also does the job. This is how it could look:
function add_canonical_link() {
global $post;
// Check if we're on a single post/page
if (is_singular()) {
$canonical_url = get_permalink($post->ID);
}
// For the homepage
elseif (is_home() || is_front_page()) {
$canonical_url = home_url('/');
}
// For category archives
elseif (is_category()) {
$canonical_url = get_category_link(get_query_var('cat'));
}
// For tag archives
elseif (is_tag()) {
$canonical_url = get_tag_link(get_query_var('tag_id'));
}
// For other archive pages
elseif (is_archive()) {
$canonical_url = get_permalink();
}
// Fallback for other pages
else {
$canonical_url = get_permalink();
}
// Output the canonical link
echo '' . "\n";
}
// Hook the function to wp_head
add_action('wp_head', 'add_canonical_link');
// Remove default canonical link
remove_action('wp_head', 'rel_canonical');The first part defines a function to define the canonical URL and create the tag to be added. With that completed, the penultimate piece of code hooks it into the wp_head part of the web page, while the last function gets rid of the default link to get avoid any duplication of output.
WordPress URL management with canonical tags and permalink simplification
29th March 2025
Recently, I have been going through the content, rewriting things where necessary. In the early days, there were some posts following diary and announcement styles that I now avoid. Some now have been moved to a more appropriate place for those, while others have been removed.
While this piece might fall into the announcement category, I am going to mix up things too. After some prevarication, I have removed dates from the addresses of entries like this after seeing some duplication. Defining canonical URL's in the page header like this does help:
<link rel="canonical" href="[URL]">
However, it becomes tricky when you have zero-filled and non-zero-filled dates going into URL's. Using the following in a .htaccess file redirects the latter to the former, which is a workaround:
RewriteRule ^([0-9]{4})/([1-9])/([0-9]{1,2})/(.*)$ /$1/0$2/$3/$4 [R=301,L]
RewriteRule ^([0-9]{4})/(0[1-9]|1[0-2])/([1-9])/(.*)$ /$1/$2/0$3/$4 [R=301,L]
The first of these lines zero-fills the month component, while the second zero-fills the day component. Here, [0-9]{4} looks for a four digit year. Then, [1-9] picks up the non-zero-filled components that need zero-prefixing. The replacements are 0$2 or 0$3 as needed.
Naturally, this needs URL rewriting to be turned on for it to work, which it does. Since my set-up is on Apache, the MOD_REWRITE module needs to be activated too. Then, your configuration needs to allow its operation. With dates removed from WordPress permalinks, I had to add the following line to redirect old addresses to new ones for the sake of search engine optimisation:
RedirectMatch 301 ^/([0-9]{4})/([0-9]{2})/([0-9]{2})/(.*)$ /$4
Here, [0-9]{4} picks up the four digit year and [0-9]{2} finds the two-digit month and day. The, (.*) is the rest of the URL that is retained as signalled by /$4 at the end. That redirects things nicely, without my having to have a line for every post on the website. Another refinement was to remove query strings from every page a visitor would see:
RewriteCond %{REQUEST_URI} !(^/wp-admin/|^/wp-login\.php$) [NC]
RewriteCond %{QUERY_STRING} .
RewriteCond %{QUERY_STRING} !(&preview=true) [NC]
RewriteRule ^(.*)$ /$1? [R=301,L]
This still allows the back end and login screens to work as before, along with post previews during the writing stage. One final note is that I am not using the default login address for the sake of added security, yet that needs to be mentioned nowhere in the .htaccess file anyway.
Claude Projects: Reusing your favourite AI prompts
28th March 2025
Some things that I do with Anthropic Claude, I end up repeating. Generating titles for pieces of text or rewriting text to make it read better are activities that happen a lot. Others would include the generation of single word previews for a piece or creating a summary.
Python or R scripts come in handy for summarisation, either for a social media post or for introduction into other content. In fact, this is how I go much of the time. Nevertheless, I found another option: using Projects in the Claude web interface.
These allow you to store a prompt that you reuse a lot in the Project Knowledge panel. Otherwise, you need to supply a title and a description too. Once completed, you just add your text in there for the AI to do the rest. Title generation and text rewriting already are set up like this, and keywords could follow. It is a great way to reuse and refine prompts that you use a lot.
An alternate approach to setting up a local Git repository with a remote GitHub connection
24th March 2025
For some reason, I ended with two versions of this at the draft stage, forcing me to compare and contrast before merging them together to produce what you find here. The inspiration was something that I encountered a while ago: getting a local repository set up in a perhaps unconventional manner.
The simpler way of working would be to set up a repo on GitHub and clone it to the local machine, yet other needs are the cause of doing things differently. In contrast, this scheme stars with initialising the local directory first using the following command after creating it with some content and navigating there:
git initThis marks the directory as a Git repository, allowing you to track changes by creating a hidden .git directory. Because security measures often require verification of directories when executing Git commands, it is best to configure a safe directory with the following command to avoid any issues:
git config --global --add safe.directory [path to directory]In the above, replace the path with your specific project directory. This ensures that Git recognises your directory as safe and authorised for operations, avoiding any messages whenever you work in there.
With that completed, it is time to add files to the staging area, which serves as an area where you can review and choose changes to be committed to the repository. Here are two commands that show different ways of accomplishing this:
git add README.md
git add .The first command stages the README.md file, preparing it for the next commit, while the second stages all files in the directory (the . is a wildcard operator that includes everything in there).
Once your files are staged, you are ready to commit them. A commit is essentially a snapshot of your changes, marking a specific point in the project's history. The following command will commit your staged changes:
git commit -m "first commit"The -m flag allows you to add a descriptive message for context; here, it is "first commit." This message helps with understanding the purpose of the commit when reviewing project history.
Before pushing your local files online, you will need to create an empty repository on GitHub using the GitHub website if you do not have one already. While still on the GitHub website, click on the Code button and copy the URL shown under the HTTPS tab that is displayed. This takes the form https://github.com/username/repository.gitand is required for running the next command in your local terminal session:
git remote add origin https://github.com/username/repository.gitThis command establishes a remote connection under the alias origin. By default, Git sets the branch name to 'master'. However, recent conventions prefer using 'main'. To rename your branch, execute:
git branch -M mainThis command will rename your current branch to 'main', aligning it with modern version control standards. Finally, you must push your changes from the local repository to the remote repository on GitHub, using the following command:
git push -u origin mainThe -u flag sets the upstream reference, meaning future push and pull operations will default to this remote branch. This last step completes the process of setting up a local repository, linking it to a remote one on GitHub, staging any changes, committing these and pushing them to the remote repository.
Enhancing focus and wellbeing by eliminating digital distractions while browsing the web
23rd March 2025
Such is the state of the world at the moment that I ration my news intake for the sake of my mental wellbeing. That also includes the content that websites present to me. Last November, I was none too please to see Perplexity showing me something unwanted on its home page. However, there appeared to be no way to turn this off, in contrast to the default page shown in a new browser tab. Then, I decided to tolerate the intrusion, only for the practice to develop over time.
Then, I happened on uBlock Origin after finding that it will block unwanted parts of web pages. While it was a bit hit-and-miss to get things going on the Perplexity website, it did the job after some trial and error. Things can change, which means the blocking may need refinement. Even so, I can handle that. When YouTube became another place where I needed to block distractions like previews of other videos during a webinar.
Now, uBlock Origin has become the only ad blocker that and I still use with Firefox. Others like Ghostery broke websites, especially that of the UK Met Office with its cookie blocking; the Ryanair one was another casualty, and became one that fell foul of Pi-hole too. Thus, they were left after me for a single shot approach. Though some websites may complain, anything that cuts out distractions has to help productivity and emotional wellbeing.
VirtualBox memory allocation error: Solving Linux Mint host issues after LLM usage
22nd March 2025
It happened to me today when I tried starting up Windows virtual machines in VirtualBox on my main Linux Mint workstation as a host after a long layover for these. They failed to start, only for these messages to appear:
Out of memory condition when allocating memory with low physical backing. (VERR_NO_LOW_MEMORY).
Result Code:
NS_ERROR_FAILURE (0x80004005)
Component:
ConsoleWrap
Interface:
IConsole {6ac83d89-6ee7-4e33-8ae6-b257b2e81be8}
Since the messages are cryptic in the circumstances, I had to seek out their meaning. The system has plenty of memory, so it could be that. Various suggestions came my way like installing the VirtualBox Extension Pack or reinstalling VirtualBox Extensions in the affected VM. The first had no effect, while the second was impossible.
However, there was one more suggestion: fragmentation of memory, much like file fragmentation on a disk drive. Thus, I opted for a reboot, which sorted out things, making it look as if that were the problem. If it comes up again, I might try compacting the memory with the following command, leaving for a while to complete due to any temporary system slowdown:
echo 1 > /proc/sys/vm/compact_memory
Because there had been some on-machine usage of an LLM, I now reckon that caused the malaise. These can be as heavy on memory as they are on processors, so fragmentation can result. That is yet another likely lesson learned from experimenting with this much-hyped technology.
How to make Firefox vertical scrollbars more visible on Windows 11
21st March 2025
While some articles on the web have reading time added to them, thus the vertical scrollbar of a web browser can act as a hint of the length of a piece. Unfortunately, they are being made less conspicuous for the sake of aesthetics and at the expense of utility. Since Firefox is the browser that I use most of the time, addressing the matter there became a priority for me, Here then is how you configure things on Windows 11.
The first step is to open a new tab before entering about:config in the URL bar and pressing the return key on your keyboard. If doing this for the first time, you will meet a warning screen that you can disable. Agreeing to the warning conveys you to the next screen, where you can enter the string "scrollbar" and use the enter key to bring up a swathe of settings.
There are two that you need to set to false by double-clicking on the pre-existing value of true: widget.windows.overlay-scrollbars.enabled and widget.non-native-theme.win.scrollbar.use-system-size. There is one more setting that you need to tweak: widget.non-native-theme.scrollbar.size.override should have a value greater than zero, the default. Using one of ten did what I wanted once I restarted Firefox. After that, I have things as I want them to be, though you may want to refine the width setting for your needs.