On a dark winter’s eve in a snow-dusted North Pole, 
Santa’s backlog of wishes was out of control. 
Getting a headache he switched on the TV, 
thinking “What might calm me down? Ah… the BBC.”. 
Fiona Bruce’s story made him sit up and take note, 
it was about Alan Turing and some programming he wrote. 
Santa now set to work with a new objective: 
“Artificial Intelligence will make my wish management effective!”. 
After hours of reading his hopes dwindled away, 
the subject is vast! Too much for one day. 

For Santa to get his backlog of wishes under control by utilising AI, what aspects of the problem does he need to consider? 

1. Digitisation: Santa receives wishes in a multitude of channels including both digital and non-digital. Any wish in a non-digital format must be scanned into a digital image from which the text can be leveraged and stored. The conversion of images to text is done by a process called Optical Character Recognition (OCR). OCR is not new, in fact early concepts date back to the 1920’s. AI image-to-text conversions utilise OCR, however AI can add further value by sense checking the output and “guessing” what text should appear in the event of impurities in the original image. 

2. Data Processing: Santa receives wishes in a free-form multi-lingual text format. While digital channels such as email and SMS are coupled with the sender contact details, the delivery address for the wish must still be extracted. All wishes must be processed to obtain the required data: Name, address, wish. Natural Language Processing (NLP) is a fusion of computer science and linguistics that focuses on the ability of computers to interpret human language. NLP emerged in the 1950’s where the process relied on rulesets to make decisions, and it inspired Alan Turing’s “The Imitation Game” experiment (a.k.a. The Turing Test, a test of a machine’s ability to exhibit intelligence in a way that would make it indistinguishable from a human). Today, AI can utilise numerous NLP techniques along with machine learning and deep learning, such that an AI can both read and write free-form text in the way that a human would (you may have heard of ChatGPT…). AI is therefore a perfect companion to assist Santa with the extraction of key data points from every wish he receives. 

3. Data Quality: Now that Santa has all wishes data stored in his chosen format it is important that he takes action to ensure the data is as accurate as possible. Validity, accuracy, and completeness are key criteria he needs to consider. For example, it’s possible that his digitisation process extracted incorrect words due to a difficult handwriting style, which then caused the data processing algorithms to translate the mistaken words into something even further away from the original text. Anomaly detection and outlier analysis could be implemented to detect data that doesn’t seem to “fit”, such as a wish requesting “a holiday in Geese”, probably meant a holiday in Wales (alright, Greece). Such methods can be implemented without AI, however leveraging them with AI can result in continued improvements in accuracy over time. Traditional methods of validation can be applied to compliment the AI evaluations, such as utilising external data sources to validate specific data points.

4. Data Analysis: Now that Santa has good quality data, he can perform analysis that will aid his Christmas preparations. He may want to categorise wishes to streamline workshop production and perform quantitative analysis to ensure supply meets demand. He could analyse delivery addresses to group presents and minimise sack-rummaging time. Route Optimisation algorithms could be applied to streamline his delivery route, considering the locations, weather conditions, and reindeer travel capacity (the distance a reindeer can fly before it gets grumpy and demands a hearty meal). While AI may not seem essential, Santa and his staff are not very technical, and even if trained, the above problems such as wish categorisation will take a lot of effort if done manually. An AI implementation could provide Santa with a more user-friendly interface and perform analysis such as wish categorisation with the huge advantage of being able to gather trillions of data points via web scraping.

5. Regulatory Compliance: Regulations such as PECR (Privacy and Electronic Communications Regulation), GDPR, and laws such as the Data Protection Act require Santa to apply appropriate security to the collection and storage of wishes and identifiable data which includes the names and contact details of children. Transparency is a key consideration when implementing AI solutions, particularly around what data will be leveraged, what it will be leveraged for, and how the results are stored and/or shared. Santa should consider informing the parents of the children of these points, as well as publishing and regularly reviewing data policy to outline the data collection, storage, and security protocols applied. Once Santa’s deliveries for the year are completed, he should consider deleting all names and low-level address detail.

6. AI Quality: This is an assessment method for evaluating an AI model across four key categories:

  1. Model Performance – The observable attributes of Santa’s business value and risk i.e. model accuracy, stability, conceptional soundness and robustness.
  2. Societal Impact – The observable attributes related to societal value and risk i.e. fairness, transparency, privacy and security.
  3. Operational Compatibility – The attributes which enable the elves to work more effectively with the assistance of any AI system, and for AI to work with any other systems that are already in place to achieve a business goal. This could include aspects such as an explanation of the model function, documentation, and collaborative capabilities.
  4. Data Quality – The attributes of a dataset which are used to build and test models and which impact model fitness. These can include missing data, data representativeness, and the quality of the production data.

AI quality and assessment is vital to ensure that models continue to perform as intended by producing reliable and expected results in a variety of real-world scenarios. Over time this ongoing assessment should highlight areas of improvement, such as model fine-tuning, that will ensure the children continue to be happy with gifts that they actually wished for.

7. Forecasting: As the years pass, forecasting methodologies can be utilised by Santa to keep him on the front foot for future deliveries. This would help him predict material requirements and ideal supply times throughout the year, as well as assisting with the resource management and training of the elves. Successful forecasting could prevent issues such as lack of materials, short staffing, skills shortages, and generally unhappy elves. AI can bring considerable benefits by being able to apply a multitude of methods, as well as selecting the appropriate method based on gift type. An AI can build a unique forecasting model for each gift in Santa’s workshop which will increase accuracy and save time. Other benefits against manual forecasting include increased prediction accuracy and lack of bias.

Can Santa afford to ignore AI? 

If Santa were to start his business today, he may be able to avoid many of the problems that he now needs assistance with. A new business where all data is collected digitally and is appropriately validated, structured, and stored might see less value in AI solutions and favour more traditional analytics. However, Santa’s business is thousands of years old and will never be in such a position. He must also consider that any purely digital solution will alienate a significant proportion of children that do not have internet access. 

Santa is fortunate that he runs a global not-for-profit monopoly. He has no fear of losing business if he doesn’t remain competitive. If a few deliveries are delayed, or a small number of children end up with holidays in Luton, his business probably won’t suffer next year. However, as we all know, if no one believes in Santa then he will cease to exist! It would be in his best interest to ensure that he keeps his performance at the highest possible level so that his key shareholders continue to believe in him. AI can help him achieve this whilst vastly reducing his workload.

With a huge sigh of relief and a booming “HO-HO-HO”,
Santa had upskilled his knowledge of AI concepts and was ready to go.
Christmas eve was drawing near, and he was prepping for me and you,
this was going to be the best technology-delivered Christmas – where all our dreams come true.

Categories:

By continuing to use the site, you agree to the use of cookies. More Information

The cookie settings on this website are set to "allow cookies" to give you the best browsing experience possible. If you continue to use this website without changing your cookie settings or you click "Accept" below then you are consenting to this.

Close