As an AI/data specialist by profession I’m lucky enough to spend time thinking about how to apply AI today to solve business problems. However it’s fun to think beyond to what the future might hold. AI will fundamentally change our life experience from entertainment to healthcare.
In 20 years the world (barring monumental disaster, thanks Putin), will look back on this era as the cusp of a great revolution. An AI driven technical revolution that opens up truely mind boggling possibilities. This article is a exploration of some of these possiblities.
Those who have worked in computer or data science for any period of time will understand that “AI” in some form has been around for a long time. Certainly for many standard machine learning (ML) algorithms, decades. However, the arrival of ever more advanced computing power and architectures in the last few years has massively accelerated what is possible and opened a glimpse of what the future will hold.
So here are five predictions that I think we will see come to pass in the next 10 years. (there is probably a black mirror episode to cover the darker possibility of all of these).
1. Movie stars will no longer need to act
The growth of deep fakes over the last five or so years has given us a peek of the future. We already can generate extremely realistic CGI, and interactive AI driven renditions of any person we would care to animate. (See the My Heritage deep nostalgia app). However it is still not 100%. We see deepfakery exposed by different camera angles, and even the best CGI human faces in 2022 are still recognisable as CGI, even if only just.
However, it is not a question of if but when. Being able to artifically render any person on screen in an indistiguishable form from the real version is a reality that will come to pass. Personal meta avatars based on our own looks will help build a huge dataset to inform AI models.
Of course acting is more than just looks. Even if we can produce a physiologically perfect virtual version of any actor there is still work to be done on the behavioural aspects. How an actor moves, their mannerisms etc. And of course the acting. As well as mapping the physical we would need to map every aspect of emotion, movement and speech delivery.
However when the point truely arrives will we need actors to act on screen or do they just begin to licence their image?
2. ai in health care will radically improve outcomes
Health systems are a huge problem globally. In wealthy nations they are hugely expensive, in poorer ones often non-existant for the most vulnerable. Many diseases are still untreatable, mistakes still happen. We are still some way from solving all of this with AI however for this post am thinking about the possible not where we are today.
Health decisioning – How to treat and allocate health care resources including drugs. I am firmly of the belief that AI will make huge differences here for a simple reason. Quite simply health decisioning is based on data, as ultimately are all human decisions. The ability of AI solutions to analyse data volumes, cross reference with additional data sources, and work in almost real time have no human parallel. Allied to patient worn sensor technology, new methods of diagnosis (sniff testing for certain ailments) etc. it is not the possibility of AI methods that is in question here, but the quality of implementation.
Healthcare management – The apallingly executed healthcare records systems, lack of joined up approaches and old fashioned beaurocracy are a massive drag on healthcare. The incentive to fix these is limited, money is just poured into the whole (the consumer pays) or drained from clinical and social care services to continue funding. Perhaps the simple end of the AI use case spectrum but the area where with appropriate expertise AI could very quickly assist. From patient record matching to process automation.
The SCIENCE (yes I’m shouting because it’s exciting) – Drug discovery, DNA profiling, personalised medicine. Again the AI superpower to analyse vast data sets is what offers so much opportunity. e.g. do certain drugs work more effectively with certain genetic makeup leading to better drug personalisation. The possiblities are endless.
For healthcare it is important to note that the human element is critical. People respond well to people but if AI allows people to be freed from specific data driven processes and decisioning they can focus on the human elements that can also have such value.
3. ai based companions will eradicate much loneliness
There is a lot of potential for damage to mental health caused by some forms of social media, typically the desire to reach artificial ideals of living a best life, or hitting unrealistic beauty standards. It’s not the technology that is the problem though it is how it is used and curated to maximise our attention and therefore ad. revenue.
The digital world can equally play a more positive role and one such area is mental health. A good example is in companionship. There are a lot of lonely people in the world whether through age, disability or social situation. We are already at a point where studies and specific use cases are showing benefit here but again it could just be a start. For Example: https://www.bbc.com/worklife/article/20200325-can-voice-technologies-using-ai-fight-elderly-loneliness
An interesting dimension here is that other studies have shown that this does not necessarily mean human based avatar companions. Some find human avatars creepy but the delight of this technology is that there does not have to be one size fits all.
4. ai will stamp out waste and fix global inequality
Arguably the biggest contradiction of the modern world is that whilst as a species we have never had more, more resources, more advanced technology, more assets and money, the distribution of this is still wildly unequal. With the will to tackle this (the incentive) plus the algorithmic aid in optimising distribution and usage (AI) there is clearly a huge opportunity to improve this situation.
Food is an example area where the role of AI will be pivotal. To use 9.5 Million tonnes of food is wasted in the UK every year. 8.4 Million people are in food poverty. (src. wrap.org.uk). How can AI play a role in tackling this? If we boil back the root causes of this waste there are many and each offer an individual challenge that could be helped by AI algorithms.
- Unsold stock going out of date – The art of inventory management is not perfect but forecasting models and consumer demand models driven by ML get iteratively better and better. Dynamic pricing models driven by sales patterns (much like the human derived reduced Aisle in the supermarket) can also work on the demand factors to tackle this. Would I be happy to save 25% on the ingredients I plan to cook tonight if it has a one day “use by date” rather than five. There is an incentive for retailers to solve for this as it directly impacts their bottom line. This does mean that retailer level waste is the smallest of the overall pie but it all helps.
- Mis-education for consumers on “best before” dates – Unfortunately the largest contributor to food wastes are households. One of the reasons is that perfectly good food is disposed of due to dates rather than actual food safety. There are moves to scrap “best before” dates entirely which could play a huge role. But how can AI help here? This is really a data challenge. If we can collate the data and use AI to understand whether there are core socia-demographic drivers of this waste, targeted education and incentivisation could play a role here. And of course using tech to match immediate need state with surplases could drive better community food sharing.
- Manufacture waste. As for retailers food manufacturers, be they farmers or food processors, hate waste as it directly hits their profitability. Both these sectors are very much early adoptors of AI. From pest control in farming, where laser guided weed zapping is already proving successful, to production line quality monitoring which quickly identifies waste generating issues, there is a good chance that AI will play a huge role in reducing this.
5. ai enabled driverless cars will massively reduce energy usage and end road fatalities
Horrifying for many driving enthusiasts the concept of driverless cars is another case of “when not if”. Whilst full driverless cars still have challenges, the addition of more and more “driver aids” continues at pace. In the interests of openness I like classic cars and think “nice cars” with good driving experience will persist. This is really about transportation.
The energy used to create motion in cars is directly proportionate to the driving style. The efficiency of the journey in terms of time taken and energy used is therefore most compromised by the fleshy lump behind the wheel. Limit the choices of the aforementioned lump and the efficiencies pile up.
We already use AI driven route finding on Sat Nav, but allied to driver aids which prioritise efficiency, energy usage can be slashed: from regenerative braking to maintaining consistent speed; ability of AI to manage and remove congestion (convoys of cars that are computer controlled could automatically join nose to bumper to form aerodynamic trains); cars can interface with the grid to charge and share energy at most efficient times for the grid etc. etc.
Safety benefits will also be huge. 90% of road accidents are widely cited to be caused by human error. Whilst fully driverless cars have some way to go they will eventually crack the technology. And then we have reached transport utopia. To enter a car in which we can ignore the driving and work or entertain ourselves, arriving at our destination in the most fast, efficient and safe manner, is perhaps one of the most exciting outcomes that is driven by AI.
If much of the above happen we are looking at a wildly improved world. And I firmly believe this is a world within our grasp. To deliver on this will fundamentally require building the human resources and skills to deliver these systems. And these are individual systems. It does not take artifical general intelligence (i.e. do it all at a human level) or artifical super intelligence (go beyond human intelligence) to acheive all these. It will just take a little more time and of course the right incentives for the right people.