Top Trends in Artficial Intelligence 2021
Most industries have faced mass disruption in 2020. One industry which has experienced continued growth is the tech industry. People are now learning to adapt to the new normal and services that automate manual processes have become invaluable. Within this the AI industry has particularly thrived with the IDC valuing the industry at £156.5 Billion a 12.3% growth since 2019.
With digital maturity being more important than ever, it is likely that the industry will continue to boom into 2021. Here are some of the trends in AI advancement you are likely to see this year.
AI for Good
With the use of AI becoming standard practice in many industries it is then expected that much of the stigma and fear of the rise of artificial intelligence is also dissipating. One industry in particular where implementation of AI is highlighting AI’s potential for good is in the healthcare industry, where its use has been vital to successful attempts in managing the pandemic.
Due to the pandemic, the health care industry has been forced to create systems to cater to different requirements then before, managing both those affected by the pandemic and patients suffering with other conditions. AI has simplified this process through the automation of processes such as managing hospital capacity and tracking vital patient data to flag the patients who are in urgent need.
One area within the automation of healthcare which has been gaining traction and we is likely to boom over 2021 is the use of robotics. Robots have been put into action in hospitals in the US this year to reduce the pressures that are being placed on healthcare workers and we are only likely to see wider application of this in the coming year.
The rapid progress in the use of AI in healthcare has showcased the potential in using AI for good and so in 2021 we are likely to continue to see AI technology being adapted to meet societal needs.
AI for Outbreak Prevention
With 2021 seeing the wide availability of a Covid-19 vaccine we are hopeful to see a global drop in cases across 2021. If we are to see the end of the pandemic It is still as vital as ever to prevent outbreaks of the virus. Key to this prevention will be the use of AI in contact tracing software. In the UK we have seen the cost of ineffective contact tracing software, with the UK treasury allocating £12 billion to the development of track & trace system. We have begun to see AI contact tracing software implemented across the world with successful uses such as in Singapore. However, while there are successful examples there is currently inconsistency in the effectiveness of implementations. Across 2021 we hope to see AI being used universally in the prevention of virus outbreaks, allowing for more effective and automated contact tracing systems which do not require people to manually scan into locations or fill out forms, similar to Singapore’s. Systems may also leverage other modern AI technologies such as computer vision, which has been tested in China to provide automated contact tracing using surveillance cameras. Widespread use of these technologies over the coming year will allow 2021 to be the year that we see the power AI possesses in prevention of virus outbreaks.
Transparency
The wider AI community has seen a rise in demand for greater transparency concerning how the technology operates. With what seems to be a constant deluge of news surrounding unethical and biased Artificial Intelligence in 2020, it is only a matter of time before government regulations are brought in to tackle the issue. In layman's terms this is the question of why an AI system makes decisions. Without transparency over the decision-making processes the technology cannot be held accountable if it encounters a fault. Artificial Intelligence is being used more and more to inform decisions making from parole hearings to environmental predictions, making any fault potentially catastrophic.
Major corporations and government agencies have tried to bring a standardised playbook to how we can create transparent AI, however they are yet to adequately deal with the issue of intellectual property. When the decision-making processes are the product it is difficult to suggest full visibility over a system. Conversations surrounding what should be considered transparent and what is protected property are likely to continue in 2021. Companies that design their technology with a mindset of full accountability will see the least disruption when inevitable regulations are brought in.
Smarter not Bigger
Artificial Intelligence’s environmental impact became a hot topic issue at the end of 2020. With Google attempting to censor Dr Timnit Gebru’s paper highlighting the carbon emissions caused by large AI systems forming the catalyst for this conversation. At Brainpool’s end of year party for our expert community a majority of our members cited GPT-3, a new language prediction tool, as the most interesting development in AI in 2020. GPT-3 has however been criticized for its massive environmental impact, reportedly the equivalent of the yearly energy consumption of 126 homes. The massive amount of energy required for Artificial Intelligence is becoming problematic.
Originally AI began in a similar vein to mobile phones that became larger the more processing power they required. These two paths diverge when smart technologies were deployed in phone hardware, allowing for smaller machines with improved optimisation. If modern AI where a mobile phone it would be the equivalent of carrying around a telephone box in your back pocket. The training systems require vast amounts raw data to be effective, resulting in massive power consumption. Methods of teaching AI are largely inefficient, forcing more complex technologies to rely on size rather than smarts. As the environmental impact ramps up it is likely we will see responses from the AI community to develop smarter and more efficient training for their models.
Written by Dominic Richmond and Joe Duszynski-Lewis