In this informative (but hopefully exciting!) blog, we look into Artificial Intelligence and the role of AI in business, as well as a brief look at AI in a nutshell and moves in AI today.

Artificial Intelligence in a nutshell:

Artificial Intelligence (AI) refers to computer systems' capability to perform tasks typically associated with human intelligence. It involves using computer science and large data sets to create machines that can solve many problems!

Techniques like machine learning, neural networks, evolutionary algorithms, and Natural Language Processing (NLP) all fall under the umbrella of AI. These methods and data sets allow computer systems to make predictions, classifications, and generalisations based on input data. Computers can respond to external stimuli based on their predictions, classifications, and generalisations. This can include interactions with humans or other AI systems, operating other machines or computers, such as autonomous vehicles, or making recommendations like suggesting a new song when you’re bored of your music!

AI Today:

In December 2022, the AI scene witnessed in awe the unveiling of the latest version of ChatGPT, a cutting-edge computational model for dialogue that can answer follow-up questions, admit its errors, challenge inaccurate assumptions, and reject inappropriate requests.

Since then, this state-of-the-art chatbot has been one of the most talked-about topics in AI. As a result, general awareness and public interest in the topic massively increased, with Google trends search steadily rising since about two years ago, with a significant spike in December 2022, with all spotlights on gpt-3.

ChatGPT is one of the many faces around the actual computational model, Generative Pre-trained Transformer 3, or GPT-3 (with rumours of GPT-4 on the horizon). ChatGPT can be used for speeding up human-related and prompt-initiated tasks, but when it comes to using the model for business tasks, the best approach would be to utilise GPT-3 directly.

The full potential of GPT-3 (and other general-purpose models) is seen in various use cases, particularly in content creation and customer support.

Examples include:

• Automated product descriptions,
• SEO optimisation,
• Shopping assistance,
• Customer support chatbots,
• Identifying customer sentiment and tone in communication channels.

Specialised AI solutions:

Specialised AI can effectively address specific problems when provided with the appropriate data and requirements. These cost-effective solutions are available across multiple business areas, making it easy to adopt cutting-edge technologies into virtually every sector!

AI can provide an enhanced customer experience and add value to your business. For example, AI-enabled solutions such as advanced product search, style-based search, AR/VR product visualisation, and virtual try-on can improve how customers interact with and imagine your products.

As well as the above, AI can assist with brick-and-mortar store operations, providing valuable insights for customer support and identifying common user complaints. Overall, AI can bring added efficiency and innovation to your business operations as processes can be more streamlined. AI helps improve customer support by providing insights into user complaints and frequently asked questions. As a result, it can assist with creating more efficient and effective support systems and identifying areas that need improvement.

Additionally, AI can help review and update FAQ sections to ensure they provide users with accurate and helpful information. Whether for ongoing operations or specific interventions, AI can bring added value to your customer support efforts.

While many AI-powered solutions can be integrated seamlessly into a user's experience, there are other applications where the technology is more prominently featured. For example, virtual Reality (VR) and Augmented Reality (AR) are advanced user interfaces using machine learning algorithms and other AI techniques. These technologies can provide immersive and interactive experiences for users and are becoming increasingly popular in various industries such as gaming, education, and retail.

AI Examples In Different Industries:

• From a logistics standpoint, AI can help manage inventory levels, optimise supply chain operations, identify trends, and forecast demand. It can also help predict and prevent potential issues from disrupting the supply chain.
• From a business perspective, AI can aid in analysing customer data and uncovering valuable insights. It can help to understand customer behaviour, identify buying patterns, and explore the success or failure of marketing and sales strategies. AI-driven business analytics can also reveal previously unseen correlations and patterns, providing businesses with a deeper understanding of their operations and allowing them to make more informed decisions.
• In the manufacturing industry, AI can help analyse order trends, detect defects and faulty goods, and identify machinery maintenance needs. By monitoring production processes, AI can help to improve efficiency and reduce costs.
• AI has also been used successfully in other areas, such as financial services, digital marketing, HR, health, and life sciences. A diverse set of methods had been established to tackle issues related to business process optimisation, trading, drug development and so much more! The possibilities are really endless when it comes to the ever-developing world of AI!

It's worth noting that AI solutions, while powerful, can sometimes produce incorrect results. Therefore, ensuring that the models are properly trained and have human oversight is vital. However, when implemented correctly, AI can save time, increase productivity and improve a business's overall performance!

What have we been doing?

At Mindera, we have trained a computational model using machine learning, the design of experiments and evolutionary algorithms to control the Android character "Flappy" and help game designers tune game mechanics. The user makes the character fly through gaps in an obstacle course.

The grounds of this work are now being tested against regular mobile applications, mimicking users' behaviour and even spotting unforeseen user conducts - exciting stuff!

How it works (the geeky bit!)

In a game like Flappy, there are only two options on each frame: either do nothing or have him flap his wings and move up a little. Apart from helping Flappy move up or down, the chosen action has no impact on anything else (let’s call it the state of information for the app).

In more complex games and applications, there are multiple actions that can be performed on each step, with even more possible outcomes, affecting the state of all the information. Furthermore, the same action may produce different outcomes depending on the current state of information. This means that just like when having a bot playing chess, the number of combinatorial possibilities rapidly increases to the point where it is not practical to iterate all paths. To help on this end, the AI logic behind Flappy’s training incorporates different strategies for figuring out which actions will perform better. And they evolve automatically in each cycle of the experiment for Flappy.

We are now using this bot capability to field test mobile apps without interfering with the application's code and in a completely “hands-off” way. We are also undergoing experiments on using this logic for training adversarial bots in a multiplayer (also with bots) shoot-em-up game!

This model approach does not depend on the business case or actual input data. For that reason it can be used in various contexts, such as simple application testing, goal-oriented testing, new feature testing, exploratory usage and performance testing with multiple user journeys at the same time, either on mobile apps, games or web applications.

It can also be beneficial to assess game mechanics changes or application changes (how users will likely behave with them). These models help in all stages of the software development lifecycle to better evaluate how users may perceive changes. It is also helpful in the exploratory tweaking of applications, usage analytics and user engagement.

Mindera’s take on AI

We want to use AI to bring the best out of your business through your people! We strongly believe that technology is an enabler for people and people-based business relationships.

“When all you have is a hammer, everything looks like a nail”.

No two companies are alike, so solutions shouldn’t be either. When crafting solutions, we go that extra mile to bring the best fit and excel customers’ expectations. As techies who solve complex problems and love contributing to the community, we take it personally to bring in the best-fit tools for the challenge. Artificial intelligence is definitely in our toolset.

We do it because we care, we wear our hearts on our sleeves, and our commitment is to deliver the best we can for you and hopefully become your true partner!

Interested in getting AI into your business strategy toolset? Let’s talk.

Does AI excite you as much as it does us?! Here are some extra sources:

• Turing, A. M. (1950). Computing machinery and intelligence. Mind, 59, 433–460. https://doi.org/10.1093/mind/LIX.236.433

• Jarrahi, Mohammad Hossein & Askay, David & Eshraghi, Ali & Smith, Preston. (2022). Artificial Intelligence and Knowledge Management: A Partnership Between Human and AI. Business Horizons. 66. https://doi.org/10.1016/j.bushor.2022.03.002 .

• Russell, Stuart J. (Stuart Jonathan), 1962-. (2022). Artificial intelligence : a modern approach 4th edition. Upper Saddle River, N.J. :Prentice Hall, https://aima.cs.berkeley.edu/

• Artificial Intelligence Journal, https://www.sciencedirect.com/journal/artificial-intelligence

This piece was jointly written by Cláudio Teixeira and Pedro Evangelista!

Cláudio Teixeira:

Account Owner, Mindera Labs

Cláudio is passionate about big challenges and making new concepts work. His main drive is on getting teams to collaboratively share the same challenge vision and goal paths. His background gives him the knowledge to understand that teamwork is all about sharing, and sharing success is the best part of it.

Pedro Evangelista

Product Owner, Data Scientist, Back-end Developer, Mindera Labs

Pedro likes to face new challenges and situations out of his comfort zone proactively. He is passionate about working with data and teams with different backgrounds to enmesh the knowledge from different sources to answer relevant business questions.