Baidu's AI predictions for 2022: Autonomous driving! Quantum computing! Space! Human-machine symbiosis!
Did a computer program tell them to write this?
Baidu Research's AI-centric "Top 10 Tech Trends in 2022" report has outlined the Middle Kingdom megacorp's predictions for technology over the coming year.
Baidu CTO Haifeng Wang describes AI as a "key driving force of innovation and development," thanks to rapidly evolving core technologies, cross-domain connectivity, and expanding applications.
It's no surprise that the list focuses on AI given Baidu's business domain. The Beijing-based company's search engine captures over 70 per cent of the Chinese market while also developing other products, particularly AI research and cloud computing. The research arm takes a deeper look at its associated technologies. Think Google but Chinese.
A similar list was produced by Alibaba's DAMO Academy earlier this month. Alibaba's business features e-commerce, cloud computing, AI, and fintech products among others. DAMO is its research arm.
Although DAMO's list features AI, it does so with less focus than Baidu Research, resulting in both overlapping predictions and completely different takes.
Not mentioned by Baidu Research but included in DAMO's list were the use of silicon photonic chips for high-speed data transmission and replacing electronic chips in computing, and soft-bodied robotics whose flexibility and enhanced perception could potentially change the course of manufacturing industry.
Three areas Baidu Research covered that Alibaba did not are autonomous driving, inclusivity, and the integration of quantum hardware and software. The pair had similar thoughts when it comes to foundation models, AI's effect on science, and privacy computing. Space exploration, humans merging with robots, and sustainability factors made both company's lists, albeit with different approaches or outlooks.
Given that Baidu operates an L4 autonomous taxi service in a 60km2 area of Beijing, it's not surprising that the list might refer to robocars as "the next generation of vehicles that will give rise to a revolution in transportation," although the company doesn't give too much detail on what that will look like.
Baidu said AI will become more inclusive, but in a convoluted way: small and medium-sized enterprises will be able to access open-source platforms to improve their businesses, thus supporting more diverse sources of innovation.
Integration of quantum hardware and software
Already becoming widely implemented, Baidu Research thinks the technology should continue to improve by increasing qubits in scale and reducing or adapting to noise.
Users will start having access to more back-end options on cloud-native quantum computing platforms, gradually proving commercial potential for the integration of quantum hardware and software.
"Government agencies, academia and the industry will collaborate more closely to build high-quality quantum devices and cultivate quantum technology talent, taking the first steps toward creating a quantum computing industry chain," Baidu Research predicted.
Large-scale pre-training models
Both the trend lists featured the self-supervised large-scale models trained on big data known as foundation models, and both predicted the world will see smaller ones as research and development shifts from attempts to increase the model size toward practical deployment.
- Meta says it's building world's largest AI supercomputer out of Nvidia, AMD chips
- IBM finally finds a private equiteer willing to purchase Watson Health
- Meta trains data2vec neural network to grok speech, images, text so it can 'understand the world'
- Nvidia pushes crowd-pleasing container support into AI Enterprise suite
"For example, built upon cross-modal large models, AI will drive the adoption of AIGC (AI Generated Content), which promises to spur creative cognition, improve content diversity and reduce production costs," said Baidu Research.
AI for science and medical
AI for science and medical uses is already pervasive. Baidu Research reckons it will further fuse data-driven and theoretical deduction and play a greater role in furthering fundamental discoveries.
In particular, the research arm expects AI-powered computational biology, spurred on by the success of mRNA vaccine technology, will be on the rise. Precision medicine will have an uptake as costs decline.
In an environment where security and compliance has become critical, thanks in part to global regulations, Baidu Research said privacy computing is not far off from development in areas like computational biology, financial analysis, and data regulations. Additionally, in the long-run, privacy computing technology may "push encrypted data circulation and computing by default," thereby creating an "infrastructure for user confidence."
Similarly, Alibaba said advanced algorithms for homomorphic encryption, which enables calculations on data without decrypting it, will hit a critical point reducing the need for computing power to support the encryption. The emergence of data trust entities to assist in sharing the sets across organizations was also mentioned.
Baidu Research also thinks AI is going to take hold in outer space, with sensor-loaded decision-making autonomous spacecraft roaming around extra-terrestrial planets carrying robotic arms, not only conducting scientific studies, but also mining moons and planets for resources, empowered by lessons learned in construction machinery automation to excavate continuously 24 hours a day. Oh yeah, and these spacebots are self-repairing as they build space laboratories and analyse big data. Really, they do everything.
Alibaba's space predictions of the future were less imaginative, focusing on the very real demand for increased satellite connectivity and how they can improve through increased interconnectivity.
With social distancing kickstarting human digital presence, and continuing to be a necessity for many for the foreseeable future, Baidu Research said humans will continue to morph into avatars and robots. Of course, AI will play a role as it infiltrates and replicates human senses and XR (extended reality) improves.
As cyborg tech becomes the norm, more platforms will want in on the market. As a result, the world will see more virtual and real mashup experiences, both in consumer situations and professional life, not only reinforcing the digital economy, but also merging with the "real" economy.
Alibaba addressed the concept of human and machine mashups, but kept it simple, with XR glasses being what drives the success of technologies such as cloud-edge computing, network communications, and digital twins.
As for going green, Baidu painted a picture of AI equipment operating in a more sustainable manner. For instance, AI processors with higher computing power and lower energy consumption will continue to be invented, but also AI technology can be applied to tasks like improving energy efficiency ratios.
Meanwhile, Alibaba's green predictions centre around how AI can help connect the dots of things like renewable energy. For instance, "intelligent scheduling using deep learning techniques could optimise scheduling policies across energy sources – like wind, solar, and hydroelectric." ®