Can you escape the AI wave?
Artificial Intelligence and technologies like machine learning, computer vision, and natural language processing are at the core of most software and IoT devices today.
While it may sound like a distant dream, filled with overly optimistic imagination to the uninitiated, artificial intelligence applications have actually already firmly set foot in many industries. Think about the iPhone face detection feature, your social media account personalized feed, and self driving cars on our roads.
In fact, the AI market size is projected to reach $1,394.30 billion according to a recent report by Fortune Business Report.
But, what does the future hold for AI?
I wondered too.
So I went out and interviewed a few AI engineers, enthusiasts, and researchers to establish what we should be excited about, what we should be concerned about, and what we should be outright scared of when it comes to AI in 2023.
In this article we are going to look at 25 interesting artificial intelligence predictions for 2023. In fact, as you’ll discover, heads are about to roll when it comes to the question of AI and ethics among other interesting predictions.
Let’s get started.
1. Rise of conversational intelligence tools
When meetings happened face to face in the physical world the information everyone had access to in a room was identical. Having an AI assistant in that setting would require things like Google glass.
But with the rise of remote work and most communication moving digital, we’re going to see a rise of conversational intelligence tools that leverage the opportunity to create uneven information scenarios.
A meeting over zoom can have a personalized experience for each individual. We can search, showcase and surface information that might be relevant to just one individual. For example in an online classroom, the instructor and student can have completely different experiences in this world.
Sourabh Bajaj, Tech Lead Google, LLC
2. Increased use of ML techniques in core business offerings
Most ML models are currently deployed to the cloud, which limits their applications. With recent advances in hardware inference engines, it is becoming possible to run machine learning algorithms in constrained compute environments.
Advances in this area will drive exciting new applications for AI, such as AR/VR and autonomous driving.
As advanced ML techniques become more accessible (e.g. with easy-to-use frameworks such as Pytorch) and industry data becomes more widely available, industries will increasingly utilize machine learning techniques in their core business offerings in 2023.
However, I think the industry will adopt these techniques in a more pragmatic method, starting with simpler techniques that are more understandable and less volatile.
Mitchell Spryn, Research Engineer Facebook, Inc
3. More government initiatives on data sharing and availability
AI for defense is hampered by two bottlenecks: accessibility and shareability.
Data is the key enabler to the development of AI, but the defense sector is a data-scarce environment as most data is restricted. Or siloed. Or unstructured.
Moreover, as the data applied to an AI-enabled system needs to be verified and assured, it is not always possible with data derived from civil sectors.
In 2023, I think we will see more initiatives from governments on sharing and availability, as governments start to recognise data as vital.
Robert Limmergard, Secretary General Swedish Security and Defense Industry Association
4. AI in the Supreme Court
In 2023, AI will see its seminal moment where the zeitgeist clashes with enterprise adoption for the future of the technology.
On one hand, we will see a high tide for environmental concerns over carbon footprints of transformer models, increased scrutiny of personal data usage, and privacy concerns over computer vision facial recognition that will lead to congressional intervention.
AI may even reach the Supreme Court. And on the other, more AI companies will submit for IPO while large enterprises will report greater earnings and cite AI adoption as a leading factor.
Money will win.
Steve Meier, Co-founder KUNGFU.AI
5. Government investments in AI research
2023 will not be a time of major technical breakthroughs but will see a significant increase in the AI market and significant resources directed to AI research and development by government and industry.
This new funding will be directed at moving AI from just data-oriented (deep) machine learning to more complicated areas including neuro-symbolic, better language models, and long-term AI planning systems (a real failure of the learning-based systems to date).
Prof James Hendler, Director Rensselaer Institute for Data Exploration and Applications
6. Deep Fakes will continue to plague the media
In 2023 we will see more democratization of AI through further releases of large pre-trained models in the Open Source community for example a big update from Open AI on GPT. This is already having a profound effect in NLP, and we’ll see that sophistication spread into end-user applications such as Chatbots.
Deep Fakes will continue to plague our media, but we can expect to see improvements in the detection and filtering of this fake content in a zero-sum game for Computer Vision models.
Finally, we are already seeing an explosion in healthcare, medical and life sciences adopting AI technologies, and due to the lengthy development and clinical trials required in these fields may see some interesting breakthroughs towards the end of 2023 and going into 2024.
Hopefully, there will also be some further regulation around ethics in AI, as we know the FDA is reviewing and introducing Good Machine Learning Practices (GMLP) which will set the foundations for stricter and much-needed regulation for AI creators.
Jack Hampson, CEO DeeperInsights.com
7. Demand for sustainability amongst policymakers
We will start to see an even greater convergence and focus on the nexus between sustainability, measurement, data and artificial intelligence.
This is being driven by greater awareness and demand for sustainability amongst investors and policymakers as well as financial institutions themselves.
To achieve the kind of alignment needed to meet global commitments like the Paris Agreement on climate, there needs to be standardization of metrics to encourage comparable, useful data.
In the future those that can combine skills and knowledge in big data, AI as well as the business, financial and scientific aspects of this issue will be very useful – this is an area where we need to start upskilling younger workers and graduates.
Matthew Chan, Head of Policy and Regulatory Affairs ASIFMA
8. Embedded ML for on-device analytics
With 250 billion microcontrollers in the world today, embedded machine learning will become the default technology for performing on-device data analytics for vision, audio, motion, and more.
In the near future, we will see this driving significant innovation from accelerating scientific discoveries to consumer devices, medical research, robotics, fully autonomous vehicles, voice-activated assistants to smart manufacturing.
For the first time, embedded machine learning will give billions of devices the brains needed to make smart decisions without needing to send data to the cloud, as it will be small enough to fit into any environment under the most constrained conditions.
Adam Benzion, CEO Edge Impulse
9. Edge computing deployments to accelerate
Building off the pandemic digital momentum, the Neal Analytics team expects to see vision, AI, and edge computing deployments accelerate in 2023.
Computer vision is well-positioned to be the new frontier, taking advantage of edge hardware developments to move closer to the point of data collection and offload AI workloads. This will open new scenarios, especially in retail and healthcare.
In manufacturing, Deep Reinforcement Learning will gain traction with more applications in production yield and supply chain optimization. Data fusion, AI operationalization, and customer order fulfillment will also become focus points across industries.
AI Team, Neal Analytics
10. AI to predict network performance
We anticipate that AI will play a big role in helping engineers understand, predict, and improve network performance.
At Energy Sciences Network (ESnet), our researchers are developing algorithms to study network patterns and identify anomalies; we’re building automated tools to study TCP traces, which helps us understand how our network behaves and develop new protocols for improving big file transfers over a WAN.
We are also using network automation tools to create intelligent dashboards that allow AI algorithms to make informed decisions by consuming multiple network statistics at once.
The prototype NetPredict Map uses advanced machine learning techniques to predict congestion points in the networks between twenty-four hours to seven days ahead of current network traffic statistics, improving real-time congestion prediction to 85% accuracy.
In addition to that, we’re also developing novel reinforcement learning approaches that could interface with network controllers to help reroute data flows to alternative paths if certain paths are busy or underperforming.
Experts at Energy Sciences Network (ESnet)
11. More AI technologies will likely get approved by the FDA
In the healthcare space, a number of additional technologies for AI in medicine will likely get approved by the FDA. Some of these technologies will likely also be able to obtain dedicated CPT/reimbursement codes.
AI in medicine will likely start to penetrate areas besides oncology, including diabetes, Alzheimer’s disease and cardiovascular disease.
2023 will also likely see more innovation in non-black-box (interpretable AI) technologies in order to make them more amenable for use by clinicians and physicians.
More innovation will continue in the use of AI not just for diagnosis of disease, but also for predicting patient outcome and predicting and monitoring therapeutic response to specific drugs like chemotherapy and checkpoint inhibitors.
Anant Madabhushi, PhD Professor of Biomedical Eng. Case Western Reserve University
12. Improved human-machine interactions
AI will be deployed with a huge impact in various industries in 2023.
There will be more and more applications where AI systems combine symbolic AI (e.g. Knowledge Graphs) with machine learning (e.g. deep learning) to incorporate context and meaning to be able to explain the results and internal processes and improve the overall accuracy.
We foresee that in 2023 AI applications will especially focus on improving human-machine interactions (e.g. chat or voice systems or human-friendly robotics) to further assist people in a more natural way in their decision makings, information search, or task automation.
Jürgen Umbrich, Senior Knowledge Scientist Onlim
13. Rise of defensive AI
Security is still top of mind for most in the infrastructure space.
As a consequence, it’s likely we’ll see applications of AI in the datacenter space not just as a tool to help humans perform Root Cause Analysis (such as the new AWS Detective service) or anomaly detection (for log files, or monitoring data) but more and more autonomous behavior where the infrastructure actively defends itself (defensive AI).
This is more and more a requirement for everybody, not just big companies because the rising use of “offensive AI” lowers the barrier of entry for bad guys and everybody is fair game today.
Alex Bordei, VP of Product Management and Engineering MetalSoft
14. Use of AI for document management
In 2023, we expect to see AI and machine learning further melded with technology.
With people working digitally and now remotely, we can expect to see more applications of machine learning and AI being integrated into everyday software, like Zoom, email platforms, or desktop document management tools.
For instance, it can be used to better anticipate end-user behavior, refine pattern recognition for email marketing campaigns, and automate smarter, more efficient document processing tasks. The possibilities are endless.
One certain thing is that we will see AI continue to play a role in developing the digital lifestyles and work environments that are merging due to the pandemic.
Reena Cruz, Brand Manager InvestinTech.com
15. Use of AI in drug discovery
Artificial intelligence (AI) and machine learning have steadily made their way into life sciences, often being utilized in areas such as diagnostics and molecule screening.
Within pharma it’s likely we’ll see these technologies being used in screening processes to speed up drug discovery and to examine how different molecules can treat different tumor types.
The vast amounts of data needed for drug development make AI and machine learning useful tools for researchers and could help result in reduced development times and ultimately bring medicines to patients sooner.
Reece Armstrong , Editor European Pharmaceutical Manufacturer
16. More AI in environmental, social and governance (ESG)
Financial institutions have been working with artificial intelligence and machine learning for years. However, in 2023, they will ramp up efforts with respect to ESG.
Responsible and sustainable investing requires listed companies, exchanges, rating agencies, fund managers, institutional investors and regulators to properly integrate huge amounts of data into risk and investment processes, along with reporting and other functions.
There’s absolutely no way this can be done without applying AI.
Jame DiBiasio, Founder Digital Finance Group
17. AI will transform more into Augmented Intelligence
I think Artificial Intelligence will start being seen more as augmented intelligence.
Instead of it being seen as a stand-alone application, AI will transform more into Augmented Intelligence, helping people in increasing efficiency in existing workflows.
For instance, take robotic process automation, or AI helping in filling a form faster by predicting the next values, based on what has already been filled. Or AI helping with writer’s block by generating ideas on what to write, or AI helping in database compression by identifying what keys are best to compress based on analyzing usage data, or AI helping in writing SQL queries.
I see the future of AI as being more of an assistant rather than stand alone applications like self driving cars or robotics.
Abhishek K, Chief AI Architect SublimeAI Inc
18. Forced government regulations on AI
In 2023 we should see more transformer based models in every field, vision transformers are going to improve but won’t replace CNNs because of their expensive nature.
Medical AI is going to increase in value, becoming one of the “hot” topics.
Countries might force some regulations on AI explainability, especially in critical systems. That might force some of the researchers to focus on XAI.
Two major players in the AI world are still the US and China. With the help of hardware manufactures we should be able to see some other types of networks on the rise (like graph neural networks).
Kemal Erdem, ML Engineer QuarkOwl LTD
19. Accelerated use of ML to optimize underlying business drivers
Machine Learning is used primarily as a prediction tool to understand what will happen.
In 2023, I expect more sophisticated companies to focus on using ML to understand the underlying business drivers they can optimize to best affect the future.
In order to gain this understanding, you need to determine the limited set of true drivers and high-order inter-combinatory effects within a large number of data columns, as well as integrate ML with an optimization engine.
An example of this is our work in Trade Promotion Optimization and Supply Chain Optimization. So moving forward, we’ll see an acceleration of focus on how ML can help drive optimal outcomes for the business.
Jason Glazier, Ph.D. CTO Enterra Solutions
20. Significant growth in the use of AI to improve traditional medical imaging
I think we will see basic AI making significant inroads into the back-office processes of healthcare, scheduling, billing, coding, and patient flow, e.g., call prompts and semi-autonomous.
Clinical applications may draw the headlines, but the only significant growth will be imaging, which is more fully digitized in a standard way. Electronic medical records and the promise of AI-fueled “Big Data” solutions will remain mired in trying to both standardize and create clean data sets.
AI businesses will continue to cater to the C-suite, who make the purchasing decisions; while continuing to ignore the end-users, physicians, nurses, and other healthcare personnel, often risking significant implementation problems if not outright rebellion.
Charles Dinerstein, Medical Director American Council on Science and Health
21. 2023 will increasingly see AI as a part of our work and life.
Artificial intelligence cuts across almost all aspects of our lives today.
Whether it is through improved machine learning algorithms that serve up more appropriate content in our streams or automated systems that manage our online conversations (chatbots) and feed our pets.
Arts and culture will gain grounds as artists continue to work with machine learning and neural network models to advance our human creativity.
Finally, 2023 will begin to solve AI’s darker side — bias and echo chambers – offering a possible utopian instead of dystopian view of the future for us all.
Brett Ashley Crawford, Ph.D. Associate Professor Carnegie Mellon University
22. AI integrated hardware will be a key area of innovation
In 2023 we will see continued innovation and adoption of AI technologies across industries.
I predict the success of AI in driving biotech, especially biopharmaceuticals, feeding off the recent success and meeting the needs of public health.
Precision farming will also be a strong area of innovation driven by continued concerns for climate and microclimate disruption.
Finally, AI integrated hardware will be a key area of innovation fueled by both high global demand and concerns on supply chain shortages.
Kevin M. Purcell, Ph.D. Professor of Data Science Harrisburg University
23. Combining of AI, IoT, and cybersecurity to create really smart infrastructure tools
All around the world countries are pushing artificial intelligence and machine learning as the key technologies to drive the advances in Smart Highway and Smart Cities.
With roads, junctions, and the transport infrastructure being covered in IoT based sensors and data collection devices that monitor traffic movements, AI is the only way to mine the masses of data, so judgements can be made based on real-world insights.
One of the major themes of 2023 is the combination of three key technologies bringing AI, IoT, and cybersecurity together to create really smart infrastructure tools.
Anthony Davis, Editor Highways.Today
24. Increase in applications with voice recognition for phone services
Many companies will use the COVID pandemic to introduce solutions based on artificial intelligence, both in the health sector (i.e. face mask video detection, virus spread models, etc.) and in the retail sector (i.e. sales predictions, opening/closing store optimization).
Due to the high increase of full/partial remote work, we will see an increase in applications that can take advantage of AI like voice detection and recognition for phone services, or text detection and understanding for digital sent of documents.
Chatbots may also increase its use for customer screening and guidance on webpages.
Santiago Morante, Ph.D. Data Science Manager at Telefónica Tech
25. PC hardware will be an integral part of AI and ML
We expect to see PC hardware continue to have an integral role in Machine Learning and AI as researchers continue to get access to next gen GPUs.
It’s incredible to think about a world where our limitations are only our own creativity and ability to innovate, but we’re close to achieving that goal.
2023 will be a very interesting year for the industry.
Josh Covington, Managing Director Velocity Micro
This was quite an interesting read.
However, I still think we have a really long way to go when it comes to making machines really think like humans.
An example on point is using AI for recruiting.
It would take an immense amount of holy grit and just being smart overall to make a machine conduct a job interview that can be deemed conclusive. Not to mention the huge datasets that you’ll need to train a model effectively.
However, there are a ton of free and open-source data science tools and artificial intelligence courses that can make this possible.
If you are really enthusiastic about learning artificial intelligence, these online courses and platforms will provide you with all the skills you need on calculus, algebra, theory and practical AI/ML application programming.
What do you think AI has for us in 2023?
Please share your thoughts below.
Lerma is our expert in online education with over a decade of experience. Specializing in e-learning and e-courses. She has reviewed several online training courses and enjoys reviewing e-learning platforms for individuals and organizations.