Using Artificial Intelligence in Astrology

AI is the shiny new object in the hands of big tech companies and if the Google CEO, tech bro, PichAI is to be believed, online users can’t get enough of its wonders. The fact of the matter is that tech conglomerates mostly see AI as a means of generating new revenue streams and cutting down on operating expenses, all in one technologically marvelous stroke. However, what presently can be said for sure about this technology is:

• It’s here to stay and will pervade into almost every major aspect of life.
• It is a powerful technology that will lead to the automation of many traditional processes.
• Those who learn how to harness its power early, will hold an advantage over those who don’t.
• Tech companies will promote each iteration as being revolutionary.
• As it grows in capability, more negative consequences will become apparent.

The planets aren’t going away, and astrology will always be relied upon regardless of technological advancements. This is why, despite all the wonders of modern technology, astrology still holds strong.

Use of Technology in Astrology

Given that the astrological community is generally open to embracing new techniques and technologies, it is thus only logical that they be early adopters of AI. The question is then, what role could AI play in the practice of astrology?

Applying AI to Astrology

At the time of writing this article, there are at least half a dozen different AI based astrology software, for the purpose of providing automated horoscope generation and predictive analytics. Most of these employ one of the following OpenAI API models:

1. gpt-3.5
2. gpt-4
3. gpt-4o
4. gpt-4.5

In simple terms, APIs are a means of submitting data to a web based programme or software, which returns new information, as processed by that software. APIs can thus be used to power new, secondary programmes.

PowerFortunes.com’s very own ‘Free Celtic Cross Tarot Readings’ and ‘Online Tarot Assistant’ are examples of programmes that run on APIs. In the case of these two programmes, we use the OpenAI API to provide interpretations of Tarot cards, as they appear, in the various positions of a ten card, Celtic Cross Tarot layout. This sort of mimics the role that a Tarot reader plays, although a reader of the cards does not simply spell out their general meanings, but does so on the nuanced basis of their position in a layout. So in essence, the instructions that are provided to OpenAI’s artificial intelligence in the above two examples, are asking it to apply its knowledge to explain what a Tarot card means when it appears at a specific location in a Celtic Cross Tarot reading. Judging by the popularity of these two services, the AI used seems to be doing a pretty reasonable job.

AI based astrology software are required to do something similar, but astrological analysis has many more aspects to it as compared to Tarot readings. Tarot readings essentially involve the cards, their position in a spread and an interpretation of the spread based on the traditional meanings of the cards, and as divined by the reader. Astrological analysis follows a more rigid and complex format, based on well defined rules and procedures. Hence the requirements on AI for Tarot interpretations are somewhat less taxing than those for astrology readings.

Will AI Replace Astrologers?

Would it be possible to replace human astrologers with those of a virtual variety? The above described example of how we use OpenAI for our online ‘Celtic Cross Tarot Readings’ and ‘Tarot Assistant’ illustrates how AI can be used to call upon its knowledge of Tarot cards and employ this to act as a virtual Tarot reader. But how well does it perform such a task? The fact of the matter is it can perform some tasks very well, particularly models like gpt-4.5. AI is good at keeping track of a multitude of information and applying its extensive database of knowledge to each piece of data thrown at it. It is also skilled at arranging all the different bits of gathered information into an accurately worded, coherent output. But does this make AI a genuine substitute for an accomplished Tarot reader or astrologer? The answer is, no and not without good reason.

There are presently only a few AI models with publicly available APIs. These models are of a generative AI type, meaning that they generate or create data/content such as code, mathematical calculations, images, videos or written content.

These generative models are LLMs or large language models. This means that their basic purpose is to generate language by predicting the most probable word to succeed another word, based on the data on which they have been trained.

The fact that LLMs are able to provide outputs related to Tarot and astrological knowledge, is because these are topics that were a part of the data on which they were trained. This is not because they possess reasoning derived from astrological principles.

Astrology predictions involve two processes, analysis and interpretation. There are twelve houses in a chart and up to twelve different planets that can occupy these houses. This means that a basic chart interpretation involves analysis of at least 144 configurations of factors, whereas a comprehensive astrology prediction will take into account many more chart related details. An astrologer will however, apply their discretion to only analyse those factors that are relevant to the prediction. In general, the greater an astrologer’s experience, the better their astrological reasoning and ability to identify those factors that are relevant to a prediction. AI isn’t designed to replicate human logic and is therefore not capable of applying this type of reasoning.

There is also another compelling reason why AI cannot replace your average astrologer. No astrologer has complete knowledge of astrology. If the mythological background to astrology is considered, then our present day knowledge of the subject is itself incomplete, as portions of the ancient scriptures on which this knowledge is based have been lost. This is explained in detail on our page about astrological belief. Therefore, while a good astrological prediction can be surprisingly accurate, it will not be perfectly accurate. To compensate for the ambiguity that a horoscope chart can throw, astrologers often make use of secondary means of assistance. This could be in the form of a different astrological method, by corroborating findings through an alternative approach, or by using a different means of divination altogether, such as psychic awareness, numerology, or the Tarot.

Effectively performing an adaptive analysis of this nature is beyond the capabilities of any present day form of AI. It is for this reason that human astrologers will not be replaced by artificial intelligence powered, virtual astrologers, any time soon.

The Advantages of Using AI in Astrology

Although the publicly available LLMs lack genuine knowledge of astrology, they do have their uses in astrological analysis. AI models are able to process a large assortment of information and are hence, well suited to processing the multitude of details that astrological analysis involves. They can therefore, potentially process more information than the average astrologer at a given time and may also be able to elicit detail that might otherwise escape an astrologer’s attention. Also, by virtue of having machine learnt much of the internet’s information, they possess a vast knowledge base. Hence AI certainly has a role to play in the advancement of astrology, in the same way that other forms of computer technology have done thus far. Some suitable applications for AI in astrology would be:

• Using LLM linguistic skills can be helpful for describing horoscope characteristics in detail. LLMs are capable of providing quick and extensive information about planetary, zodiac and house related traits and can potentially even add new ways of describing astrological phenomena. They can for instance, be used to recount all possible influences that a planet may have on an area of life, relating to the house over which the planet rules. This is possible through the systematic analysis of the house and sign in which the planet is aligned.
• Similarly, AIs are able to highlight all the different possibilities that arise from the conjunctions and placements of different planets. However, not all of the highlighted information is likely to be relevant or even accurate, hence human oversight will always be necessary.
• It is quite possible that during the course of a reading, an astrologer may struggle to recall all the intricate details of the astrological phenomena observed. Therefore, input from AI can prove to be valuable for clarifying insights about general astrological belief, and historical and mythological background. An astrologer can thus potentially use AI to cross check their conclusions. This is helpful in ensuring that no detail is overlooked.
• Machines have a definite advantage when it comes to quickly analysing large amounts of data. This is of help in astrology readings where factors such as past and future planetary positions, periods and transits need to be correlated.

The Disadvantages of Using AI in Astrology

It is important that both seekers of astrology readings, and astrologers themselves, avoid getting caught up in all the hype over AI capabilities. It is probably apparent by now that to assume that this technology can independently conjure up precise and meaningful predictions, would be naive. An AI based astrology software will require the chosen model to provide an analysis in accordance with a fixed set of instructions. These instructions have to be carefully designed, so that the model understands what is expected in each step of the process to achieve the final result. Some method of self appraisal may also be used, to keep the AI on track, however errors of the following nature are still likely:

• In order to get specific details about an alignment or condition of a house, complex and detailed instructions will invariably be required. Although some models are certainly better than others, prompts with multiple steps and conditions, are likely to trip up most present models. Having said that, some are better than others. In our experience, of those presently available, the OpenAI gpt-4.5 model is decidedly better than most, although still far from perfect. Some users do suggest the opposite however, when this model is used for other purposes, such as for content creation. Either way, gpt-4.5 is still, quite slow and expensive.
• All AI models can inherit and amplify biases of their training data, leading to misleading or completely untrue results. In astrological terms, this is often reflected by stereo typical characterisations of zodiac sign and planetary qualities. That is, if left to its own devices, an AI will be unable to discern nuances, or curb its instinct to refer to an astrological factor, in a preconceived way.
• Once an AI latches on to a certain line of thinking, it is not easy to get it to change its approach. This can be mitigated to some extent, by careful prompting. However, if a user wants to provide an AI model with a standard prompt, for the unattended analysis of similar data sets, a model is liable to get stuck in a mode that will differ from the desired outcome.
• The characteristics of a horoscope chart can often be contradictory. For example, one planet may indicate a certain type of affliction, while another planet may mitigate this condition. A good astrologer will be able to apply their experience and knowledge to weigh up such conflicting factors, to find out where the truth actually lies. An AI model is not capable of applying this type of discretion, and is likely to err on the side of generalisation when presented with such a dichotomy.
• AI hallucinations are well documented and information about astrology is no exception to a robot’s tendency to make up facts. What seems to trouble present day artificial intelligence is information overload. That is, if an AI is presented with the intricate details of a typical astrology chart, it tends to go off on its own tangent, while analysing factors. This is obviously less than desirable and can take some effort, or even require a complete reset of the chat, to bring it back on track.

Final Thoughts

AIs certainly have a role to play as virtual astrological assistants. However despite all the claims by AI companies, allowing them to fly solo and make astrological predictions without any form of human intervention, would at the present time, be unwise and ill-advised.

Having said that, this certainly is not the end of the story, as far as AIs are concerned. AI will have a role to play in all aspects of modern life in the future, astrology included. Developments in this area are extremely fast paced and what is valid today, won’t be next year or even next month. In the context of astrology predictions, as explained in our earlier article on the subject, ‘Guide for Using ChatGPT for Astrology’, a model that is trained specifically on astrological data, principles and previously successful predictions from authoritative sources, is what would be required to create a true astrology based AI. It is only a question of when, and not if this becomes a reality. However, even if this were to be achieved, it is difficult to categorically say whether even this type of, astrology trained model would be successful at making independent, but accurate predictions.