After a year of AI

I’ve been dabbling with Perplexity AI for nearly a year now and decision time is looming. Should I renew my subscription. Or should I look elsewhere (e.g. Halomate)? Or should I swear off AI altogether?

At the moment, I feel no need to look elsewhere. I try to be circumspect in my use of AI, so I have not attempted to exploit the full potential of Perplexity. Therefore, in spite of occasionally being tempted by the latest bells and whistles offered elsewhere, I see no point in starting again with something else.

The question of whether I should abandon AI is less straightforward to answer. There is a real case to be made against its use. As far as I can see, that case boils down to two arguments:

• AI encourages mental laziness. We are tempted to accept what it tells us as true without critical engagement. Often, the output is so polished (at least superficially) that we are tempted to submit it as our own work – hence the deluge of AI slop on the Internet.
• The environmental argument. AI data centres put extreme demands on power and water supplies. They run so hot that they raise temperatures in their locales measurably above what would be expected (at the same time, overheating rivers with waste cooling water). Some have massive backup diesel generators which produce great quantities of air pollution. And these ills are growing exponentially with the AI bubble. On top of this, there is the psychological harm to the humans who are employed (often on low wages in developing countries) to monitor and censor inappropriate outputs.

I’m not terribly bothered about the threat of mental laziness. The crucial point is always to treat the output of AI in the same way as you treat any other secondary source – always insist on references (and always check them) and never plagiarize. In any case, unlike many chatbots, Perplexity AI is essentially an AI-powered search engine and that is how I use it (as a Google search on crystal meth).

But I remain conflicted on the environmental and wider social issues. Yes, AI speeds up my research at a time of life when I may not have very much time left. But does the improved efficiency of my research justify my implicit support for the damage that AI is doing on a global scale? If I were a rugged individualist like so many tech bros, the answer would be an unequivocal ‘yes’. But I’m an Orthodox Christian, so I have to balance the goods of AI (not just my vanishingly small contribution to the sum of human knowledge buts its role in pushing forward the bounds of our knowledge in so many fields) against the environmental and social ills. I’m unable to come to a clear view on this, so I suspect that I will continue to use Perplexity for the time being.

Blackwell Idealist

I have lost count of the number of computers I have used over the years. But there has been one constant factor: I have installed Idealist on every computer I have used since the mid 1990s.

Originally developed by Blackwell Scientific in the early 1990s, it was originally marketed as ‘the Information Manager’. According to the manual, ‘It is designed mainly to manage textual information, and is particularly suited to information that cannot be organized into a strict and regular structure.’ It is a simple and very flexible textual database, which uses an index card metaphor. Essentially, it is a computerized card index system. And because each entry looks like an index card, it presents a familiar, friendly face to old-fashioned academics like me.

The version I am currently using is the January 1999 build of Idealist 3, which was issued just after Blackwell sold the program to Bekon. Unfortunately, the new owners’ ambitions rather outstripped their technical abilities. Version 4 was dangerously unstable (the only program I have ever known that completely trashed Windows when I tried to install it!). They did eventually bring out a version 5, which ironed out most of the bugs in version 4 and introduced rich text formatting, but it failed to catch on and the program disappeared from sale in the early 2000s.

Strengths

• Simple and rapid. Each index card is an individual record in the system, and everything is indexed automatically unless it is defined as non-indexable (which can be done by creating a ‘stopword’ list or by excluding particular fields from the search). As a result, searches are virtually instantaneous even with very large databases.
• Extremely flexible. One of the strengths of Idealist is that you can have different record types in the same database. So record types for articles can be stored, sorted and searched alongside those for books, chapters, quotes, manuscripts, maps, computer programs, and so on. This means that you can store your data in records designed for the data rather than trying to fit your data to a rigid record structure.
• Utterly reliable. To give you some idea of what I mean, I have used it on every version of Windows from Windows 3.11 for Workgroups through to the 64-bit version of Windows 11 Professional (and for several months I also used it on Linux Mint with the help of Wine) and, unlike some well-known programs I could mention, it has never crashed and has never lost any data.
• Effectively has no limits regarding size of database. Originally the size of the database was limited by the amount of RAM available on the computer, which could be an issue in the days when the size of RAM was measured in megabytes. But it is a very small program and the 32Gb of RAM on my laptop would have no difficulty in handling an Idealist database far larger than the largest one I currently have (which contains about 25,000 records). Each database can hold up to a million records and each record can be up to 8Mb (though there is a 64K or 10,000 word limit on the length of an individual field and a maximum of 160 fields per record).
• Easily manipulable. For example, unlike most databases you can create new fields within a record or redefine record or field types on the fly. Users can define their own record structures, alter the ‘stopword’ list, or define search synonyms.
• Powerful search facilities. Its search ability is very well thought out, enabling you to drill down quickly to precisely the records you need. In addition, simply highlighting a word or phrase in a record and pressing the ‘Stack’ button will take you to a new hit list containing all the records in which that word or phrase appears.

Weaknesses

• Plain text only. Sadly, Idealist works exclusively with ANSI plain text files. If you want to use rich text, or foreign characters, or store other kinds of data (e.g. images), you have to look elsewhere. In an ideal world, I would like a database that allows me to store Unicode text files and equations in LaTeX.
• Poor linkage between records or to other files. It is not straightforward to create such links and they have to be updated manually if any changes are made to those records or files.
• Limited export capabilities. You can export from Idealist, but it is not straightforward and the options are fairly limited. Personally, I tend to limit exports to Idealist natural files, which are essentially structured plain text files that can be edited in Word.
• Loss of functionality on 64-bit systems. Idealist was one of the first 32-bit programs available on the market. Unfortunately, it comes wrapped up in a 16-bit set-up program, which, of course, will not work on 64-bit systems. There is a simple workaround, which is why I am still able to use it. (And one of the members of the Idealist users’ group has recently created a 32-bit set-up program for it. See discussion here.) In addition, some commands no longer work properly (I’m not sure why this is, but again I found that you can work around the problem by direct editing of some of the program files).

Idealist and my workflow

The main reason I like it so much (apart from its utter reliability over nearly a quarter of a century of use) is that the card index interface fits perfectly into my workflow. My approach to research is a fairly simple-minded process of analysis followed by synthesis, and Idealist is the cornerstone of the analytical stage of the process.

As I read, I capture individual ideas, quotes, etc. that seem relevant to what I am researching. Each gets its own entry (with title, reference and associated notes/comments) in my Idealist Notes database and full bibliographical details go into the associated References database. The result is a vast soup of ideas, which I can search almost instantly.

The synthetic stage is usually a process of mind map construction during which I draw on repeated searches of Idealist. Sometimes I prepare the mind maps on paper, but more often these days (and certainly with more complex projects) I use XMind.

The end result of the mind mapping is an outline that I can work up into an article or book. In the past, I have usually done this in a general purpose word processor (originally WordPerfect but more recently Word, not because I think it’s better but merely because most of my editorial clients required me to use it). However, I am currently experimenting with Scrivener for Windows for the final writing stage. And, of course, Idealist is still essential at this stage as I keep referring back to it for quotes, snippets of text and references.

How to win a Nobel Prize

Paul Krugman, winner of the 2008 Nobel Prize for Economics, has some interesting advice on his website for anyone embarking on a piece of research, which can be summarized thus:

• Listen to the Gentiles: Read outside your discipline. In his own words, ‘Pay attention to what intelligent people are saying, even if they do not have your customs or speak your analytical language.’
• Question the question: The questions asked in any academic discipline are theory laden. There is no such thing as a bare fact in any subject. Everything is affected by the presuppositions of the discipline, so Krugman rightly advocates paying critical attention to those presuppositions.
• Dare to be silly: Don’t be content with the safe and the familiar. Dare to strike out into uncharted territory. My favourite example is that of Richard Feynman: the research that led to his Nobel Prize was inspired by asking why a dinner plate thrown across a dining room at Cornell University wobbled as it flew!
• Simplify, simplify: Keep shaving with Ockham’s razor. Personally, I prefer A.N. Whitehead’s approach: ‘seek simplicity, but distrust it’. Or, to paraphrase Einstein, theories should be as simple as possible, but no simpler.