AI Predictions for 2021 – What The Experts Say

Artificial intelligence (AI) grew rapidly in 2020. This was not only due to the digital transformation but also the global pandemic which forced businesses and individuals to remain at home, rather than their offices, for most of the year. Nonetheless, 2020 accelerated AI usage and development.

And it does not look like AI usage and development is going to slow down. We spoke to 20 different experts in the industry to hear their predictions for the coming year and what we can expect for 2021.

 

Our Panel of Experts:

• Jonathan Kriegel – CEO of DOCOMO Digital
• Gareth Cummings – CTO of eDesk
• Dr Alex Young – CEO and Founder of Virti
• Jonathan Crane – Chief Commercial Officer at Amelia
• Lukas Kinigadner – CEO of Anyline
• Dr. Elina Naydenova – CEO and Co-Founder of Feebris    
• Shane Lowe – CEO and Co-Founder of Vitrue
• Eran Kinsbruner – Chief Evangelist, Perfecto by Perforce Software
• Nick Ford – Chief Technology Evangelist at Mendix
• David Newns – Founder and Chairperson of Prevayl
• Tom Ollerton – Founder of Automated Creative
• Ifigeneia Mavridou – Affect Scientist and Researcher at emteq labs
• Rudy Benfredj – CEO and Co-Founder of mendelian
• Chris Puttick – CTO/COO at ContrAI
• Simon Menashy – Partner at MMC Ventures
• Andy Green – Chief Information Security Officer of Gemserv
• Bela Stepanova – VP of Product at Iterable
• Bill Admans – Senior Vice President of Operations for OWNZONES
• Tai Alegbe (CEO and Co-Founder) and Rajpal Wilkhu (CTO and Co-Founder) – Contingent
• Einat Metzer – CEO and Co-Founder at Emedgene

 

 

For any questions, comments or features, please contact us directly.

 

 

Jonathan Kriegel – CEO of DOCOMO Digital

 

 

“Deployment of convolutional neural networks in the drug discovery and trial data assimilation in humanity’s fight against the COVID-19 pandemic has awakened the world to the mainstream applications of artificial intelligence that will only accelerate in 2021 and beyond.

Elon Musk backed Open-AI’s Dall-E comes just a few months after OpenAI announced it had built a text generator called GPT-3 (Generative Pre-training), which is also underpinned by a neural network. The language-generation tool is capable of producing human-like text on demand and it became relatively famous for an AI program when people realized it could write its own poetry, news articles and short stories.

Algorithmic Commerce

Visual search will see massive growth. For example, a partnership between Snapchat and Amazon, called Visual Search, allows you to point their app-camera at an object, then get a link showing either the product itself or something similar, available for purchase.

Pinterest, meanwhile, has a multitude of visual search tools, such as Shop the Look, which uses machine learning to “dot” every object in a photo. Google takes this one step further with the Google Lens app, a general visual search engine. It does more than just identify products for sale—it decodes an entire landscape.

And IKEA has taken things the farthest. By using their AR app via smartphone, you can map your living room into a digital version with exact dimensions. Need a new coffee table? Their technology lets you try out different styles and sizes. It will even allow you now to preview multiple pieces of furniture at once. Your choice triggers a smart payment, and just like that, an Ikea customized coffee table is delivered at your doorstep. Need help assembling? Their AR app can walk you through it step-by-step. Social commerce powered by AI algorithms will be another area of growth.

The recent partnership between TikTok and Shopify will allow merchants to create native, sharable content that turns their products into in-feed videos that resonate with the communities within the platform.”

 

Gareth Cummings – CTO of eDesk

 

 

 

Dr Alex Young – CEO and Founder of Virti

 

 

“AI is increasingly being used in the medical sector, which is a really exciting prospect. It has helped us to tackle the virus by analyzing and interpreting data on the virus’s spread. It is also being used in healthcare to help with treatment and medical training. With AI and machine learning being the ultimate problem-solving tools (tools which are only getting smarter), we can expect to see the use of AI in medicine continue to grow over the coming months and years.

As we continue to live our lives remotely, AI is also being used to enable businesses to train and develop teams at a distance. At Virti we’re helping a range of different companies – from medical device manufacturers to hospital staff and big corporates – use AI and Virtual Reality to deliver effective training at a distance and deliver enhanced, data-driven feedback to learners.

Finally, businesses will increasingly use AI to bring behavioural analytics and personalisation into their operations. Whatever you’re selling – be it clothes, shoes, food or software – understanding your customers’ behaviour and tailoring how you reach them accordingly is going to be essential. We’re doing more online than ever before, so digital competition across all sectors is tough. Now is the time for businesses to start leveraging the rich data that’s available to them using AI.”

 

 

For any questions, comments or features, please contact us directly.

 

 

Jonathan Crane – Chief Commercial Officer at Amelia

 

 

“2020 has seen intelligent AI and automation supporting professional roles in a number of ways. In insurance and banking, for example, AI-enabled digital colleagues have started to work with customer service representatives as ‘whisper agents’. These colleagues are not only able to source the information their human counterparts need to reply to individual customer queries, but they can also understand the industry’s complex regulations and provide recommended responses in line with them.

The potential for AI to release mundane admin hours across all types of white-collar roles in this way is massive. This not only presents a huge opportunity to improve business productivity, with more time concentrated on value added tasks, but will be a key enabler in allowing employers to reduce the burden on employees.

We’re likely to see this continue into 2021 – with the explosion of AI-powered customer service in healthcare too. With higher demand on public services, intelligent digital colleagues present a cost-effective and reliable solution for healthcare providers looking to scale operations. The world will be more focused on disease management in general after 2020, and as a result there will likely be more pressure on healthcare professionals managing routine, day-to-day check-ups for patients alongside the fallout from Covid-19. The opportunity for conversational AI to manage this burden is huge – allowing professionals to focus on delivering care where they are best served to do so.”

 

Lukas Kinigadner – CEO of Anyline

 

 

“Although vaccines are on the way, 2021 will start much as the last year ended, with our lives restricted in many ways due to health-related restrictions. This has driven the huge demand for practical AI-powered technologies that enable ‘contactless’ experiences, and this trend will only continue to grow this year.

For example, as travel opens up, we will see more hotels enabling guests to complete their check-in and receive their room key right on their own smartphone by scanning their ID with their phone camera, avoiding the need for a front desk. ‘Self-shopping’ will also become more common as retailers give their own customers the ability to scan product barcodes, price tags and even expiry dates, then pay through their own devices.

Until fairly recently, integrating AI technology for business uses has been mostly limited to companies with extensive R&D departments. However, the rise of low- and no-code platforms like Mendix will significantly lower the barrier of entry in 2021, making AI tools far more accessible to businesses in a shorter time frame.

Speaking on a more technical level, one area where will see big steps forward this year is in the pre-training of AI, through which networks can be optimized on specific data sets, and annotation can be entirely automated. This will save hundreds of man-hours for machine learning engineers, and lead to far faster time-to-market for AI products.

In this regard, an area of development I’m particularly interested in is ‘few shot learning’, the method by which neural networks can be trained on a very limited set of data points rather than the thousands currently required. One of the ‘fathers of deep learning’ Sepp Hochreiter is leading a project on few and one-shot learning at his AI Lab in Linz, Austria that I believe could completely reimagine how computers ‘learn to learn’, and would again create huge optimizations for how computer vision tools are developed.”

 

Dr. Elina Naydenova – CEO and Co-Founder of Feebris    

 

 

“We will see significant advances in the development and use of AI in healthcare, particularly in community settings as the NHS transitions to more integrated care at a time when frontline care workers are dealing with unprecedented challenges. Advanced algorithms are already augmenting clinical decision-making in some hospitals, it’s time they were used to address workforce challenges in the community.

At Feebris are working relentlessly to deliver the next generation of remote monitoring, powered by AI, with the aim of ensuring that everyone around the world has access to early diagnosis and treatment. Working collaboratively with clinicians, social care and public health professionals, we have developed and calibrated AI-powered technology to ensure generalisability across clinical and community settings. 

The pandemic has highlighted that change is urgent. I hope that we continue to see AI-innovators explore and develop new functionalities that enable greater usability and accuracy of decision-support for multi-disciplinary health and social care teams.”

 

 

For any questions, comments or features, please contact us directly.

 

 

Shane Lowe – CEO and Co-Founder of Vitrue

 

 

“In the healthcare sector, the use of AI is sure to see continued exponential growth in 2021. From rapid diagnosis to disease management and drug delivery, AI will deliver unprecedented improvement to global medical care. But it’s not just clinical settings where AI will be changing the game; this year will see the general public continue to embrace AI in their homes to both monitor and improve their health.

The events of the past year have made us, as a society, wake up to the importance of prioritising our mental and physical and wellbeing, and consumers are on the lookout for the best new tech to support their endeavours.

Because AI technology has the ability to capture vast numbers of data points simultaneously, to analyse large data sets and ‘learn’ at an incredible rate, AI tech can deliver the targeted and specific health advice that today’s consumers are looking for. Using simple apps or web-based tools, it’s now possible to self-check your skin for signs of cancer, to optimise your desk setup to prevent musculoskeletal injuries, to monitor your post-operation recovery and much more.

The rapid pace of innovation means that the costs of using AI tech are no longer prohibitive, driving a democratisation in access to personalised healthcare. The potential for continued growth in this space is very exciting, and definitely an area to watch.”

 

Eran Kinsbruner – Chief Evangelist, Perfecto by Perforce Software

 

 

“From better digital advisors to financial customers, through to better end-user operations, such virtual assistants, AI and ML technologies will grow into their next level. There are already organisations, such as banks, using NLP to better serve the banking customers and advise on next steps like investments, savings and more.

In the world of insurance, AI and ML will become key in streamlining and automating the claims management, submission, as well as processing. Bots developed by insurance companies will query their new customers to tailor a policy for them (health, car insurance, others). Based on history and massive data management, those firms will optimise and fine tune their pricing and offering to their end users.

When it comes to fraud, AI and ML — through huge data processing, smart algorithms, correlation to patterns, automated identification of anomalies  — will become more sophisticated and boost security management. As mobile and digital are the main channels for many customer-facing services, security must be enhanced, because mobile devices vary across devices, operating systems, geographies, and with new ways of app consumption (app clips, APKs). Companies must ensure that such services are 100% secured, at all times. Here, AI and ML such as Force Fraud Detection from Shift, can be useful.

AI and ML in 2021 and beyond will be used by organisations throughout the entire digital services development and monitoring. There will be a spike in building smarter virtual assistants that can automate engagement with the users (NLP, Bots). In addition, there will be an increase in big-data analysis to allow businesses to improve service availability, the functionality that is used by users, prevent outages, reduce time to restore services that are down, and more.”

 

Nick Ford – Chief Technology Evangelist at Mendix

 

 

“During the pandemic, the true value of AI has been realised as businesses across the world needed to rapidly respond to the pandemic. AI quickly came to the fore by enhancing tasks related to workforce planning, simulation modelling and demand projection, helping to accelerate tasks faster than ever before. For enterprise businesses, its use has been helping companies transition to remote working – which for many had to be done overnight.

This year we will continue to see exponential growth throughout 2021 in the UK AI sector, but the biggest shift we can expect will be that of AI-embedded tools becoming more widely adopted, especially in the enterprise as businesses turn to AI for analysis, recommendations and to increase employee productivity.

Although AI-enabled decision-making is already taking place in physical settings, whether it’s applying for a loan or getting a speeding ticket – this year we will see more sophisticated applications of artificial intelligence and machine learning across industries. As part of this shift, a greater focus will be placed on integrating models and capabilities into multiple business processes and operations to drive improved insights and data collection.

Further to this, research and development teams as well as software developers will make greater strides into developing the ways AI can be made more accessible to citizen developers by operationalising AI using low-code technology. This will revolutionise AI as we know it making it more accessible by removing any barriers to entry the technology possesses. Not to mention, this would also provide the opportunity for citizen developers to leverage AI to solve business problems without the need for a developer or data scientist.”

 

 

For any questions, comments or features, please contact us directly.

 

 

David Newns – Founder and Chairperson of Prevayl

 

 

“I think machine learning has such an unbelievable opportunity to help humanity. Computers can make decisions way faster than we can, they can come to conclusions that we humans are reaching the limit of being able to do at the same speed. So we have to really harness the power of computing to crunch mass data points and actually bring them together to give us data that can help us automate positive choices. It’s a huge opportunity.

Take healthcare for example. The global medical community is seeking to answer how companies like mine can get these health data points from people and use them to predict the sorts of illnesses and diseases they get, and then work with healthcare pros to develop interventions. That’s what the team at the company I co-founded, Prevayl, is doing.

We’re giving the mass market fashion-first smart clothing with incredible technology that was previously only available to elite athletes, to improve their health and wellbeing. Our new smart clothing can read highly accurate data and turn it into precise insights and actions. Machine learning is a key part of it. My predictions about how it will evolve this year are as follows.

1. Clothes will give AI more human data points than watches – I think what’s exciting about smart clothing is, compared to a smart watch, we get more abundant data from the body, in more depth and at a greater accuracy than any other kind of wearable technology. And this is important because the human body is so nuanced. Small differences in your heart rate, core temperature or heart rate variability can actually be significant.
2. AI that knows when you need to take a break – physically AND mentally – A good example of data from clothing that a watch cannot offer is strain and recovery. Recovery can be hugely underestimated. We’re able to give people a recovery score, which takes in all sorts of data points in the body. So the insight might be that you know, today isn’t a good day to work out really hard because the data is showing that you need more recovery otherwise you could do more harm than good. This means you can now make a truly informed call on how to proceed, in this particular case about switching to an active recovery programme instead of something more intense. So the accuracy of data is absolutely crucial to ensure the machine learning insights and advice are precise. That’s just one example, but it gives you an idea about how this can begin a new era of preemptive healthcare.
3. AI will help extend life – With this information, at scale, we can then give people all kinds of better lifestyle choices. On the one hand you can extend life and on the other make society better able to afford its globally aging population. So therefore machine learning can massively help with the likes of healthcare, globally. It’s incredibly exciting.”

 

Tom Ollerton – Founder of Automated Creative

 

 

 

Rudy Benfredj – CEO and Co-Founder of mendelian

 

 

“The use of AI and technology in general will continue to play an important role in healthcare, as the system has had to accelerate its adoption of new ways of working to help combat the covid crisis. We will see collaboration in the ethical use of data to underpin the power of AI.

At the end of last year we launched our partnership with Modality NHS, to form the biggest ever primary care focussed rare disease diagnosis programme in the UK. The partnership utilises MendelScan, our core solution, at scale and allows us to capture the data that will demonstrate the true impact of digital health for rare and hard to diagnose patients. This gives us a wealth of opportunities for further learning and insights.

Rare diseases have been deprioritised due to the COVID19 virus understandably being the central and critical focus for much of the NHS and care system, and it looks as though the backlog of non-urgent care will continue well into this year. 

We hope that our work can help alleviate some pressure on the NHS. Our specially developed technology identifies the signs, symptoms and features of potential rare diseases in patients electronic health records to help the clinicians to reach a diagnosis. This means patients can receive quicker access to the correct treatment, care and support”

 

 

For any questions, comments or features, please contact us directly.

 

 

Chris Puttick – CTO/COO at ContrAI

 

 

“My predictions for AI in 2021 include that we’ll continue to see more AI-assisted startups and more AI and AI-powered services being adopted by established companies.

We’ll also see continued hype about “generalised AI” while the real progress and most value continues to comes from niche AI, applying existing AI technology to new areas with little or no modification to the AI engine itself.

Linked with this, 2021 will see increased numbers of startups that are seeking investment claiming they have an AI-based USP that is basically just a neat decision tree built in a spreadsheet. Kinda AI but very early 1990s, probably not even expert system level.

As a consequence, there will be a rise in the number of investors criticising startups for claiming AI is used when it isn’t – but basing that criticism on thinking only a generalised AI is AI, aka “they are just using machine learning”. Everyone’s “just” using machine learning.

In addition, I predict that this year will see more people realising data is everything for a high value AI. If a startup using AI as a core provider of its USPs isn’t pitching for investment with a primary strand of “get more data” then their AI might have some clever innovations in tweaks or application, but it’s going to get eaten by more data-hungry competitors.

Related to this, more UK and European investors will realise, as the A16Z team noted last year https://a16z.com/2020/02/16/the-new-business-of-ai-and-how-its-different-from-traditional-software/, that an AI-centric company is not like one that is software nor SaaS, even if they are also software or SaaS or both.

An AI-first company, product or service needs to spend a lot of money upfront and adopt a data-centric early years strategy. Money needs to be spent acquiring customers not because, as per traditional SaaS companies and products, that’s how you drive up gross margins, but because data is the moat; it drives the improved effectiveness and functionality of the service so increases value to the user.”

 

Simon Menashy – Partner at MMC Ventures

 

 

“AI themes for 2021: AI faces get creative: You might have seen very realistic, AI-generated human faces before. Often with clickbait headlines like “none of these people actually exist”. While impressive, they’ve largely been static photos, or simple animations. In 2021 we will see these faces get creative, appearing in everything from quirky videos to TV ads. Expect at least one ‘deepfake’ crisis where a politician or celebrity gets into trouble for something dreamed up by a mischievous programmer.

The data ops boom continues: 2020 was an astounding year for cloud infrastructure and data ops businesses. Data warehousing company Snowflake beat Airbnb to the title of top IPO, raising $3.9bn in September and soaring to a current market cap of over $80 billion. Segment, another hot IPO candidate in the cloud infrastructure space, was snapped up by Twilio in October before it went public. Big players like IBM, Google, HPE and VMware also made large acquisitions.

Entrepreneurs and VCs have spotted the opportunity and capital is now flowing into the space. Our own portfolio has kicked off the year strongly, with behavioural data platform Snowplow announcing a $10m funding round last week. Expect to see much more.

Every company is a data company: As a VC, we are constantly pitched by entrepreneurs claiming to have the latest, smartest AI technology. A couple of years ago we published research showing that 40% of supposed “AI companies” actually use little or no AI. Before AI, it was ‘big data’.

We predict that this year the pendulum will swing back, and every startup will be a data company – claiming to have unique datasets, smart personalisation models, and the highest percentage accuracy in their models (without explaining “90% of what?”). We’re looking forward to another year of fun sorting out who is really walking the walk!

Generative AI: Chatbots with friendly names may have been all the rage a few years ago, but they haven’t lived up to the hype. Instead, this year we’ll see a new wave of language generators creating content almost indistinguishable from human writing. Cutting-edge AI technologies like GPT-3 (from Elon Musk’s OpenAI) will be used for everything from creative fiction to sales emails – and your children’s homework!”

 

Andy Green – Chief Information Security Officer of Gemserv

 

 

“Our prediction for Artificial Intelligence (AI) in 2021 is focused on how AI can impact cyber security and how can it be used to further enhance technologies across the cyber security sector.

Cyber security-related AI predominantly falls into two key areas:

1. i) AI-powered cyber security controls defending our systems, infrastructures and information; and
2. ii) How and where AI will be used by cyber criminals to conduct cyber attacks to gain access to our systems, infrastructures and information.

Looking ahead to possible trends in 2021, we predict that significant development into AI-powered cyber security controls will be a key trend this year, whilst an attack that is publicly attributed to AI is likely further afield.

With regards to security control areas that will most likely benefit from greater AI adoption and integration, it is predicted that it will most likely be the controls that can benefit from predictions based on very large data sets, such as Security Incident and Event Management (SIEM) Solutions, Threat Intelligence and controls combatting disinformation including deep fakes.

Looking more closely, AI in SIEM solutions can aid more predictive capabilities through machine learning, understanding and recognising malicious behaviour warnings beyond its initial data input. Similarly, incorporating AI with threat intelligence allows for increased efficiency in processing unstructured information, ensuring more rapid response times, and subsequently minimising the risk of security incidents. Finally, AI can be used to combat disinformation by learning to detect deepfakes, and adapt to new deepfake techniques as they surface even when human detection is difficult.”

 

 

For any questions, comments or features, please contact us directly.

 

 

Bela Stepanova – VP of Product at Iterable

 

 

“In 2021, artificial intelligence will allow marketers to not only deliver campaigns at the right time, but also to create the right messages for millions of people based on their individual needs. AI technologies will allow businesses to understand how customers feel about their brands and products on a deep level, and predict customer needs proactively.

The data speaks for itself. According to a recent survey of B2C marketers by Iterable, 88% of marketers have made AI a key part of their 2021 marketing strategy and plan to expand their in-house AI capabilities in the new year.

AI technologies will generate actionable insights about customer needs, allowing brands to communicate with empathy at scale. Leveraging AI to personalize content and timing of messaging will be key to nurturing and engaging customers through the remaining challenges of the pandemic and into an economic recovery.”

 

Bill Admans – Senior Vice President of Operations for OWNZONES

 

 

“Our first prediction for 2021 is content owners will adopt artificial intelligence (AI) and machine learning (ML) for managing cloud-based workflows and media libraries. Traditionally, media files have been stored using on-premises infrastructure with little innovation around technology or the way media has been managed. The shift to working remotely has brought a new focus on using the cloud to boost efficiency and drive down costs.

We are seeing interest in our award-winning FrameDNA and Deep Analysis services within the OWNZONES Connect™ platform which utilises AI and ML to scan entire libraries and identify the core components within media files. The services are used to enrich media with descriptive metadata to drive business processes around content discovery and monetisation. The AI behind Deep Analysis identifies duplicate media for removal from a library.

OWNZONES also employs AI to create virtual masters for alternate versions and localised content. Together these services can reduce the storage footprint of content libraries by as much as 70%, which equates to millions of dollars saved.

We also predict more AI-based automation of business processes as content moves into the cloud. Media asset management is rapidly moving to the cloud along with the business process systems content owners use to monetise media. The OWNZONES Connect platform is fully extensible meaning every aspect of our toolset and workflow is accessible. We seamlessly connect to a myriad of systems used to orchestrate automated workflows generated from sales orders.

Content rights management systems connected to our platform automatically update contracts and regional availability as new media is generated within the platform. Metrics derived about every aspect of how media is manipulated and stored within our platform are used to generate reports for cost tracking and billing. New sales orders can trigger OWNZONES Connect to rapidly encode, package, and deliver without human intervention, vastly accelerating the time to market of content.”

 

Tai Alegbe (CEO and Co-Founder) and Rajpal Wilkhu (CTO and Co-Founder) – Contingent

 

 

“At Contingent we’re committed to helping our customers make better decisions faster, AI plays a crucial role in that goal. We believe AI can enhance our core search and proactive monitoring solution.

AI-based machines can predict future supply chain problems and disruptions, at Contingent we look forward to continuing to leverage these technologies in the future.

Three areas where we see key developments in AI in 2021 are in NLP, MLOps and IoT:

• On the NLP front, Open AI will make GPT3 available to all, enabling more research projects kicking off and use of the Transformer model and Deep Learning in wider applications. They will announce the release of GPT4 which will be a significantly deeper and larger neural network over GTP3. Microsoft and Google will follow with more pre-trained models to compete within the transformer models. NLP and the transformer model will become the norm in identifying risk from unstructured data – a key to our unique offering.
• MLOps will be the standard as is serverless technology to implement AI/ML within cloud native apps. As such, tuning deep learning models or hosting your own AI/ML pipeline from scratch will not be common and will be replaced with transfer learning.
• IoT devices with GPUs such as Nvidia’s kits will enable AI in IoT devices and will be introduced into home automation – Smarter doorbells, smart mirrors, etc.

In a wider context, the industry where we will see the most advances in AI will be Health Care: Diabetes – insulin management, cancer treatment, pipelines to identify health issues ahead of time (eg predicting diabetes, Parkinson’s disease and even cancer), embedded monitoring devices for various illnesses or even discovery of drugs through accelerated clinical trials.”

 

Einat Metzer – CEO and Co-Founder at Emedgene

 

 

Faster development to clinical adoption in healthcare

“If there’s one thing we learned from Covid-19, it’s that the biomedical community can come together to rapidly solve problems, building on existing research and initiating fast testing and implementation protocols. I predict that we’ll be rapidly innovating across many healthcare areas now that we’ve proved this is possible. For example, there are roughly 300 million patients with rare diseases that are undergoing a diagnostic odyssey that is rapidly becoming a healthcare crisis. With genetic sequencing costs reduced dramatically, and interpretation of genetic data solved via AI tools like Emedgene’s, we can as a community, rapidly apply our resources to solving this problem.

Accelerated AI adoption in healthcare

2020 saw an increasing number of health organizations evaluating AI tools. Increasingly AI models will cross the barrier from R&D to daily clinical use. In large part, this is because health organizations have learned to evaluate AI tools and safely integrate them with human oversight. The addition of explainable AI models will accelerate this trend.”

 

 

For any questions, comments or features, please contact us directly.