Pricing your AI training data project

Lou Salmen 28 Feb 2023

Artificial intelligence (AI) is only as good as the data it’s trained on. Put bluntly – garbage in, garbage out.

To ensure more accurate, immersive and engaging AI experiences, AI engines must be trained on large volumes of high-quality data. But preparing the AI data you need to train your machine learning (ML) model is a monumental task that can consume up to 80% of AI project time, leaving little time to focus on developing, deploying and evaluating your AI applications. One possible solution? Working with the right AI data partner to deliver the exact AI training data you need.

But how much does AI data cost?

AI data vendors take different approaches to pricing AI training data. Some vendors price hourly based on actual time spent preparing the data, some price based on the number of data points delivered, while others price based on productivity considering the time it takes to complete each data task and the total number of data tasks required.

Regardless of pricing approach, the cost of AI data ultimately depends on four key components:

People
Productivity
Process
Place

People

When budgeting for AI training data, the people you want to collect, annotate/label or validate your AI data can have a significant impact on the cost of your data. Below are the people factors that impact AI training data pricing:

Number of participants – what is the total number of participants you want to collect, annotate, or validate data?
Geographic, demographic, sociographic, or physiologic requirements – do you need participants to have specific features e.g. come from certain countries or regions, belong to a particular age range or ethnicity, speck specific languages or dialects with certain accents, have a particular skin tone etc.?
Specialized skills – does your data task require a resource to have specialized knowledge e.g. multilingual skills, computer programming skills, legal expertise, medical doctors, specific hobbies, etc. or can anyone perform the task?

Productivity

Another key component that must be considered when budgeting for your AI training data is task productivity. The following task productivity factors will affect your AI data budget:

Data type – what type of data e.g. text, audio/speech, image, or video, has to be collected, annotated or validated?
Number of data points – how many data points do you require from each participant? A data point is one output from one participant such as one image, one video clip, one audio utterance, etc.
Time to complete a data point – how long would it take a participant to perform one data point from start to finish excluding setup and training time?

Process

The process used to perform your AI data collection, data annotation and data validation tasks also influences the cost of AI training data. The following procedural factors play an important role in AI data pricing:

Training – do the data collection, annotation, or validation tasks require project-specific training prior to completion?
Reasoning – do the tasks require reasoning where the answer is not explicitly mentioned in the data type but the pieces of information can be used to deduce an answer.

Example 1 (no reasoning): Label all names within the sentence:

“Mary met Sally for a walk on the beach.”

Example 2 (reasoning): Answer the question:

“What is the capital of France?”

Moderation – do you require a moderator to guide and oversee participants as they perform project tasks?
Objectionable data – will resources be required to view objectionable data or content e.g. graphic violence, explicit language, etc.?
Personally identifiable information (PII) – do you require participants' personal information such as names, full face photos, geographic or contact information, etc. to be identifiable in the data?

Place

The setting in which data tasks must be performed can also greatly impact AI data pricing. Below are a few criteria related to place that must be considered:

On-site vs. remote – do you require resources to be on-site or can they perform tasks remotely? If on-site, does the project need to be performed in a certified facility?
Environment – does a specific environment/scenario have to be set up?
Equipment – is specific equipment such as cameras, recording devices, or props that a typical individual is not likely to have access to required?

AI Data Budget Worksheet

Each of the criteria outlined above impacts your AI training data budget differently. Some have greater impact than others, making budgeting for AI data a complex undertaking.

To simplify the AI data budgeting process for you, the RWS TrainAI data services team has developed an AI Data Budget Worksheet to help you understand:

Important criteria to consider when planning your AI training data needs
The approximate budget level required for your AI data project (answer 8-10 questions)
The impact of different criteria on your AI data budget

For data you can depend on, choose TrainAI by RWS – your responsible AI partner. Click here to learn more.

Lou Salmen

Strategy and Development Manager, TrainAI

Lou is responsible for strategy and development within RWS's TrainAI team, working closely with clients to ensure their AI projects exceed expectations.

All from Lou Salmen