Accessibility and AI in Online Trading: Opportunities and Barriers for Brokers

Brokers play a crucial role in democratizing access to trading, ensuring inclusive trading platforms that cater to a wider audience, including those with disabilities.

A notable trend Devexperts has witnessed is the exploration of AI usage for accessibility. 

With AI rapidly transforming various sectors, could it be a solution to provide meaningful progress for accessibility in the trading space? What are the opportunities for brokers who are willing to adopt this approach? 

In this article, we unpack accessible trading platforms to provide a one-stop detailed overview for brokers. 

Topics we will cover: 

• The driving force behind development
• Why AI matters for brokers 
• Examples of AI and accessibility in websites and applications
• Near-future possibilities in AI for brokers to consider
• A balanced and realistic standpoint on AI and accessibility 

Driving force of accessibility development

To begin, let’s explore the main drivers of increased accessibility development. The introduction of new technology has made designing accessible software much more feasible. This combined with changes in legislation to ensure equal access for a wider audience, has created an observable shift in the importance given to accessibility in online trading. 

Here are a few examples of catalysts: 

Web Accessibility Guidelines (WCAG)

The World Wide Web Consortium (W3C) established international accessibility guidelines and standards to ensure everyone can use websites and applications, regardless of ability.

Local legal requirements 

Accessibility is increasingly becoming a legal requirement in many regions. Legislation like the Americans with Disabilities Act (ADA) and the European Accessibility Act (EAA) makes digital accessibility a legal obligation for software companies. This transformation shifts accessibility from a competitive advantage to a necessity.

Assistive technologies

Tools like screen readers, voice recognition software, screen magnifiers, and refreshable Braille displays have helped users interact with technology independently.

Why AI accessibility matters for brokers

Accessibility in online trading isn’t only about making things easier for disabled people. It’s really about how easy it is to use a product, regardless of whether someone has a disability or is short on time. 

To function effectively, AI needs to provide accurate data, describe the world around it, and deliver essential features in a comprehensible manner. This not only eases the burden for people with disabilities but enhances usability for everyone.

Accessibility covers a wide range of impairments, including:

• Visual: Blindness, low vision, and color blindness.
• Motion: Difficulty using a mouse or keyboard.
• Auditory: Full or partial deafness.
• Speech impairment: Inability to speak or use synthetic speech.
• Learning: Dyslexia, ADHD.
• Cognitive: Trouble understanding information.

Examples of AI and accessibility in websites and applications

Structured and semantic

Websites adhering to accessibility rules utilize clear, logical markup that facilitates content processing by neural networks. Semantic HTML tags (like <header>, <nav>, <article>, and <footer>) help neural networks understand the context and relevance of different page elements. 

This improves data analysis and classification, so AI models can interpret and extract information more accurately.

Speech-to-text technology

Modern advancements in speech-to-text technology, such as Whisper, have significantly improved the accuracy and usability of voice controls. 

At Devexperts, we utilize these advancements in our AI chatbot, Devexa, to process and analyze user messages. From reading news and describing charts to reporting portfolio metrics like balance and buying power, AI is revolutionizing accessibility in trading terminals. 

Improved text recognition

Alternative text (alt-tags) for images enhances accessibility not only for individuals with visual impairments but also for neural networks analyzing visual content.

Clean and readable code

Websites built with clean, readable code that follow established standards enhance the ability of parsers and neural networks to effectively analyze web pages minimizing errors and improving data interpretation.

Screen reader support

Websites optimized for screen readers provide essential information for training AI models engaged in speech synthesis and recognition. Metadata and ARIA tags enhance the preliminary work on the synthesis of speech and comprehension of text.

Metadata and micro-markup

The use of metadata and micro-markup (such as Schema.org) helps neural networks understand the structure of content. This additional data provides more context and helps neural networks extract the right information. Micro-markup also helps improve search results and content recommendations.

Near-future possibilities in AI for brokers to consider

There is a popular notion that neural networks will one day fully understand user interfaces and adapt them to individual needs. 

AI and GenAI applications will change how customers interact with data.

For example, a retrieval-augmented generation for trading data is being developed for brokers and dealers that will make it easier to manipulate and analyze trading data.

In addition, AI-based applications will help detect suspicious behavior by analyzing data. This integration will enable dynamic solutions for complex tasks such as investment research and enhancing fraud detection continuously adapting to the evolving challenges of AI and the financial industry.

Here is a list of advancements expected to shape AI capabilities, that are potential opportunities for forward-thinking brokers:

Multi-modality

AI systems will manage a broader range of use cases developing a deeper understanding of the real world. 

Icon Interpretation

Another promising application of AI is in interpreting and explaining the meaning of icons, even when they lack proper accessibility tags. 

Accessibility tags are elements or attributes added to the code to ensure that web applications and software are accessible to all users, including those with visual, auditory, or cognitive disabilities.

This approach could reduce the burden on developers to add text labels and invest in accessibility, also enabling countless legacy software systems to become more accessible.

Accessibility overlays—Plug-and-Play solutions

Proponents of AI envision a future where user-side neural networks are intuitive enough to describe any interface and adapt it on the fly to the user’s needs. Already in the industry, we are seeing commercial offerings called “accessibility overlays”. 

These solutions claim to be turnkey, automatically analyzing and modifying web pages to meet accessibility standards. For brokers, these overlays can simplify the process of making platforms accessible, demonstrating their efforts to democratize trading and support investors.

Customization and adaptability

Future AI models will move away from one-size-fits-all approaches and become increasingly tailored to specific applications. This shift will increase effectiveness in niche areas to better align with user needs.

Proprietary data

The success of AI systems will increasingly depend on access to proprietary data. For example, companies like Apple are well-positioned due to their rich applications.

Ensemble models

Instead of one expert model, the future will see the rise of ensemble models—collections of specialized models that work together. One model might analyze structured data, another interpret visual information. A master model synthesizes their outputs for integrated insights.

Hybrid data access 

Combining semantic search with structured data will improve accessibility. The hybrid approach will allow users to navigate and retrieve information more seamlessly.

Autonomous generative AI

Generative AI (GenAI) applications are closely monitored to avoid errors and are typically used for internal tasks. In the future it will be used more broadly, creating new opportunities for customers and new uses of AI.

Retrieval-augmented generation (RAG)

This technology will become common allowing teams to manage their own knowledge bases. These systems will let users control who can access what information and how AI-driven assistants respond.

A balanced and realistic standpoint 

While we have explored the potential of AI for accessible trading platforms, it’s also important we provide you with a balanced view. 

AI neigh-sayers argue that AI won’t become intelligent enough to perform some predicted functions accurately. 

They also point out that neural networks, a method in AI that teaches computers to process data in a way that is inspired by human thought processing, need extensive training data. Given the current challenges in accessing data, there isn’t enough quality data available to explore that route for the time being. 

As a result, AI often can’t fully describe inaccessible interfaces. Let’s take the case of accessibility overlays as an example. The reality is that they often fail to solve accessibility issues and can even make them worse. When AI handles accessibility, developers lose control over the outcome. 

Here is a list of issues with AI and accessibility in online trading. 

Inability to replace traditional mechanics

AI accessibility can’t replace traditional accessibility mechanisms. And since accessibility is needed as much as possible, the transition to neural networks can’t be at the expense of traditional technologies. This means we must either avoid using neural networks or we adopt a dual approach.

Legal compliance 

In developed countries, there are already laws in place that regulate accessibility, including aspects of technical implementation.

Right now, there are obligations that make the developer of the original interface and traditional mechanisms (such as system accessibility APIs) more responsible for accessibility. 

So, if you need to be certified and comply with the standards, you don’t have a choice but to use traditional technology.

Loss of designer control

If a neural network misinterprets data designers may lose control over the user experience.

Performance varies across applications

Neural networks are all about learning, so they work better on more typical or commercially viable applications. As a result, their performances vary considerably. 

For example, there’s a widely spread model for recognizing images but not for recognizing mathematical formulas and graphics.

Brokerage solutions with built-in accessibility features

At Devexperts, we’re dedicated to making our solutions for brokers and banks accessible to everyone. We have added new features that address different user needs. 

Here are some of the key initiatives that we’ve implemented in our products:

Keyboard navigation

We’ve integrated circled modal window keyboard navigation to all our interactive elements so users can navigate easily.

Interface scaling

We’ve also made some changes to the way the interface scales. Our apps now have scaling features for Android and iOS so users can double the size of the interface for better readability.

Screen reader support

We make sure that the labels we use are clear and precise so that screen readers can effectively communicate the function of elements. 

This helps users understand what each element does. For example, we label a search function “search” rather than “magnifying glass”. We also show you how many tabs you have open, so you can keep track of them. 

The interface scaling feature for Android and iOS allows users to double its size for better readability.

We use AI to help out with things like generating ALT text for images, providing real-time captions, simplifying complex articles into more digestible formats, and even writing accessible code. 

However, we want to be clear that AI is only a supporting tool. It doesn’t have the same empathy and human judgment as we do, so we monitor this closely. 

The final verdict

There’s no doubt that AI is creating new opportunities for brokers to improve the accessibility of their offerings. 

Improved online trading for disabled users is obviously important for democratizing trading, and it’s also an opportunity to foster solid client satisfaction. 

AI is a difficult dance for brokers (and firms across many industries)—there is so much potential up for grabs, with equal uncertainty about the pace and direction of evolution. It helps to lean on experts with experience. We are experts, and we’re always ready to help. Get in touch with our team.