Will AI set a new standard for green coffee quality?

  • The SCA’s cupping guidelines were first introduced in 1999
  • Scoring coffee in this way is subjective, as humans carry inherent and unavoidable biases
  • AI could offer a truly impartial solution for assessing green coffee quality, which could have real benefits for producers

COFFEE QUALITY is inherently subjective. This means that there has been a longstanding conversation about how to objectively define it. And as pricing mechanisms rely on a quantifiable measurement of quality, this represents an issue that is challenging to resolve. 

In recent years, standardised systems like the SCA Arabica Cupping Form have been developed to assess coffee quality. These methods score features like acidity, balance, body, and flavour profile to determine cup quality and overall value.

The SCA cupping protocol and others like it have become universally recognised across the industry.

However, while these tools have been honed over decades, they fail to address a problem at the heart of human-led sensory assessment: individual bias. The protocols are structurally sound – but no matter how hard we try to be impartial, human involvement in the process cannot be unbiased, and as such, we cannot reach a definitive assessment of “quality”.

But the risk this poses isn’t just limited to the cupping table.

For example, while there are longstanding trends to work towards, subjectivity may cause a gap between what farmers and buyers recognise as being high-quality. Additionally, it is often the buyers that are in control of these scoring systems, and are therefore more easily able to influence the final sale price.

Additionally, taste preferences vary across the world. This fact is what has led some in the industry to re-work the Coffee Flavor Wheel to align with regional understandings of flavour – as the language surrounding flavour and quality in the coffee industry often fails to consider the experiences of people in the Global South. 

To resolve this issue, stakeholders in the industry are increasingly looking to technology. On a long enough timeline, this could remove the human element from the process altogether – and drive a completely new approach to evaluating green coffee quality. 

AI could bring greater objectivity to assessing green coffee quality

Could AI remove subjectivity in sensory assessment?

Across the coffee industry, automation continues to drive conversations about reducing inefficiencies and removing human error. Predictive roasting technology, for instance, is one major example of this. But could artificial intelligence (AI) do the same for sensory assessment?”

AI is no longer a “thing of the future”. It’s being used in personal and professional contexts all around the world; OpenAI’s ChatGPT has over 100 million monthly users; with over a billion visits over the same time period. It was the fastest-growing application in history until Threads surpassed it in July this year.

Essentially, AI is everywhere – and the coffee industry could well harness it to drive the industry forward.

“AI can be seen as a technological tool to work alongside coffee companies to reduce the quality differentiation between buyers and sellers, thereby reducing industry inefficiencies,” says Nicholas Yamada, business development associate director (LATAM) at ProfilePrint. “This is especially relevant due to the role quality has in trading coffee.”

Nicholas explains that AI can quickly predict the quality of coffee by converting complex molecular data from coffee samples into digital fingerprints – which has big implications for how coffee is traded.

“Currently, each country has its own grading standards, which requires traders to be familiarised and calibrated with other countries’ standards,” says Nicholas. “As a certain coffee moves up the supply chain, there are multiple quality checks that can disrupt the flow of the supply chain if these standards are not met.”

“Using AI, we can predict the cup score of green beans without the need to roast and cup it, considerably reducing the time spent in the whole process, and the subjectiveness that can currently impact a final grade.”

A clearer understanding of “quality”

To the average coffee drinker, the idea of “digital fingerprints” and AI quality checks may seem like a distant concern. While this may be the case, the implications for consumers are very real.

There are a lot of inefficiencies in the coffee trade and with how coffee is priced,” says Nicholas. “With the implementation of AI, I believe both buyers and sellers will be able to trade better and faster, therefore, pricing will be adjusted to reward both sides in this new competitive landscape.”

There are also significant potential benefits for coffee producers, who have long struggled to assess the quality of their coffees in a way that aligns with market demand. Nicholas explains that using AI won’t look to redefine how coffee is graded – rather, it aims to provide producers with a better understanding of the quality of their product.

“By having a better understanding of what a certain cup score means from the perspective of the market, the idea of what ‘quality’ is would be much clearer than before, allowing producers to focus on replicating these learnings harvest after harvest.”

Despite the potential benefits, some worry that AI quality grading could make Q graders and other green coffee sourcing professionals redundant. This is a trend that extends well beyond the coffee industry, however – of 1,000 businesses surveyed by Techopedia, 93% said they planned to expand the use of ChatGPT, and 48% had already begun replacing their workers.

However, in some cases, AI is designed to work alongside people. “During harvest season, we are able to efficiently assess and qualify large volumes of beans (with each scan taking just seconds) – freeing up time for human experts to focus on just the high-quality beans that have been identified, or on other valuable tasks,” Nicholas says.

One potential pitfall of relying on technology to assess coffee quality is its inability to keep up with the continuous innovations within the industry, and the shifting perceptions that have made “quality” so hard for humans to define. For example, 40 years ago, the majority of people preferred their coffee to be dark and intense; nowadays, many consumers want bright, sweet, and even acidic flavours. However, this is what makes AI so spectacular (and, for some, scary) – its ability to keep up.

“AI evolves at the same pace as the market does,” says Nicholas. “Therefore, it can be easily adjusted to meet new or enhanced quality checks at any given time.”

Ultimately, AI stands as an efficient and unbiased adjudicator, capable of adapting to what the market decides “quality” means. For producers, it presents an opportunity to truly understand how to deliver quality coffee, and get paid properly for it. For buyers, however, it may take some convincing to move on from the score sheets.

Coffee Intelligence

Want to read more articles like this? Sign up for our newsletter here.