Uses of artificial intelligence, otherwise known as AI, are becoming more common. Although, new uses are starting to occur that we could never predict they would benefit us in. Uses for AI have moved away solely from the production of technology and computers and found a place in forensics, agriculture, and even beer brewing1,2,3.
One potential future use for artificial intelligence is in the world of forensics. Footprints are an important part of forensic science1. They are one of the most common types of evidence and can help to reconstruct events and create profiles on suspects1. There are very few specialists left capable of easily identifying footwear and draw their own conclusions from them1. Information on this subject in the forensic field is found on the footwear database1. Those that are less specialised in using the footwear database and drawing conclusions from this type of evidence are the ones that can benefit from an AI system1. This has been achieved through using a second neural network, that has been used to identify the make and model of shoes1. It does this by analysing the impressions left by the footwear and identifying the different components of the footwear1. Examples of aspects they identify include logos or writings on the shoe and the type of tread in the shoe1. These different components are given a specific code, which they use to search the footwear database1. The AI not only provides codes, but it will also indicate if there is a chance the code may be incorrect, so it can provide the person with options for them to look into3. This allows the process to still have a level of human control, as they recognise there can still be errors, and this is made to assist and not replace human effort.
This study also compared the success rates of whether correct analyses would occur, between casual users of the database, the AI they had built, and the forensic footwear experts1. The footwear experts had the highest success rate, at almost 100%1. The AI was the second most successful, as it was right between 60-91% of the time1. The least successful were the casual users of the footwear database, as they were only able to get it right between 22% and 83% each time1. This indicates that while the AI used was not perfect or as successful as the human experts could be, it still had a higher success rate than those not used to the system and who had less forensic footwear experience. The main issue with this process at the moment is that it is hard for the AI to account for natural wear, which will change the appearance of the shoe, and make it harder to identify1. So, while this method shows promise, and can have important implications, there are still some shortcomings that mean that experts can’t be replaced.
Another unexpected use for artificial intelligence is for the process of pollination2. Using AI, it has become possible for robots to pollinate plants2. These robots would be used by farmers, and provide benefits such as more efficient pollination services thus producing a higher yield of crops2. Being able to still have efficient pollination occurring has become a greater concern as more pollinators such as bees are affected by pesticides and climate change2. As well as this, particularly in greenhouses, the act of pollination is typically performed by commercially reared bumblebees3. These bumblebees are not always effective, and this is mainly due to location, as they won’t work best under all environmental conditions3. Sometimes the use of bumblebees in some countries is banned, such as in Australia, resulting in some plants having to be pollinated by hand3. AI controlled pollinating robots can work to help fix and assist with this issue.
A robot has been designed to identify flowers that are ready for pollination3. Instead of the insects using buzz pollination, the robots use air pulses that copy the behaviour that would have been performed by bumblebees3. The robot can identify the flowers that require pollination through the use of AI3. As bumblebees are not allowed in Australia, the process of pollination in the tomato industry is performed by humans, who will touch the plants with a vibrating pollination wand every couple of days3. This process takes a lot of time and labour3. By replacing this with AI robots, this can save on a lot of time and costs3. By using this AI assisted robot, it can help make sure that effective pollination still occurs when bees and pollinators are being negatively impacted, and also in situations where the pollination in the past had to be done by hand.
Lastly, who would of thought that AI could be used for brewing beer?4. A company called IntelligentX have started creating beers from a recipe that has been altered by machine learning (machine learning is a subset of AI)4. This company produces 4 different types of beer4. After people purchase and drink the beer, they are welcome to provide feedback, which is what machine learning uses as data4. The AI can provide advice based on this data, which the company in charge can decide whether they want to use or not4. The company is then able to use the AI to make informed decisions based on customer feedback4. This means that the company can utilise large amounts of data to create the perfect beer.
In Australia, researchers have created a robot called RoboBEEER, to create consistent pouring of a beer, and to specifically create a consistent amount of foam4. This has been done as enjoyment of beer has been tied to the amount of foam poured into a beer4. Some of the components of the beer poured they can track include bubble size and beer colour4. This was tracked by taking videos4. The feedback from these videos by participants as well as their reactions during the videos was analysed by AI4. This information was given to a neural network, which allowed for an 80% success rate of prediction on whether someone liked the height of the beer4. On a broader note, they were also able to predict whether people would like a beer as a whole, and they were right 90% of the time4.
So, there are many new and exciting ways that AI is being used now, that we may never have predicted when AI was first introduced. These indicate the large variety of impacts AI can have on our lives, and how important they could be moving into the future. As well as this, this indicates the potential that AI has, and that we aren’t even close yet to seeing all the avenues that AI could be used in.
 Matthew Robert Bennett & Marcin Budka. The Conversation. We trained AI to recognise footprints, but it won’t replace forensic experts yet. (August 2021). https://theconversation.com/we-trained-ai-to-recognise-footprints-but-it-wont-replace-forensic-experts-yet-161686
 Mike Cherney. The Wall Street Journal. BUZZ OFF, BEES. POLLINATION ROBOTS ARE HERE. (July 2021). https://www.wsj.com/articles/buzz-off-bees-pollination-robots-are-here-11625673660#:~:text=The%20robots%20were%20developed%20by,the%20flowers%20to%20pollinate%20them.
 Matt Brann. ABC News. Pollinating robot looking to shake up $900m greenhouse tomato industry. (July 2020). https://www.abc.net.au/news/rural/2020-07-08/pollinating-robot-trial-starts-in-australian-tomato-greenhouse/12429244
 Bernard Marr. Forbes. How Artificial Intelligence Is Used To Make Beer. (February 2019). https://www.forbes.com/sites/bernardmarr/2019/02/01/how-artificial-intelligence-is-used-to-make-beer/?sh=24f6a9d070cf