IBM’s premier conference, Think, took place in Orlando last week. Many of the week’s biggest product announcements started life in IBM Research labs.
Last week in Orlando, Florida, IBM hosted Think, its annual premier conference for leaders across business and technology to come together to find solutions to their biggest challenges. At the conference, IBM unveiled a slew of new technologies, many of which had their genesis in IBM Research. Perhaps the biggest of those was unveiling of watsonx.ai, IBM’s pathway to the latest AI tools and technologies on the market today. And many of the models and development studio inside watsonx.ai were designed or built by IBM researchers, and were on the show floor of Think this year.
Here’s a complete rundown of everything that the team at IBM Research showed off at this year’s Think:
Announced as part of CEO Arvind Krishna’s keynote at Think, watsonx.ai is powered by a curated library of IBM’s trusted foundation models and is complemented with the latest and greatest open-source models, including many from Hugging Face. You can experiment with, tune, and deploy generative AI software with ease to a range of enterprise issues.
If you’re interested in joining the waitlist to get access to the watsonx.ai Tech Preview Program, click here. You can read more about watsonx.ai here — or watch IBM SVP and Research Director Darío Gil break down what’s in watsonx.ai in his keynote from Think below:
Earlier this year we partnered with NASA to create a suite of geospatial foundation models, which are trained on remote sensing data, like satellite imagery, that NASA makes available to the public. The first suite of models we built, called Prithvi, was trained on the Harmonized Landsat Sentinel satellite dataset, which captures surface reflectance data across the globe at a 30-meter resolution.
The model, part of IBM’s watsonx.ai geospatial offering, is planned to be available in preview to IBM clients through (EIS) IBM Environmental Intelligence Suite during the second half of 2023. It could help estimate climate-related risks to crops, buildings, and other infrastructure, valuing and monitoring forests for carbon-offset programs, and developing predictive models to help enterprises create strategies to mitigate and adapt to disasters from climate change, such as flooding and wildfires.
On the show floor, IBM researchers showed off an interface for visualizing geospatial data. Read more about the work here.
By the end of this decade, we believe practical quantum computers could start to impact how companies think about their computing strategies. Quantum computers will also profoundly alter how we secure everything online. Traditional cryptographic methods could prove easily beatable for future quantum systems. Now is the time to start thinking about how to protect your enterprise from quantum threats.
At IBM Research, along with constantly pushing the state-of-the-art in quantum computing, we’re also working to ensure that systems will be secure from potential future quantum threats. And at this year’s Think, we announced our quantum-safe roadmap, our plan for how we plan to use technology to ensure companies have the cybersecurity capabilities required for the future of quantum computing. Here, the team showed off a demo of the IBM Quantum Safe technology. It’s a set of tools designed to make your organization quantum safe. Some of the tools in the kit include IBM Quantum Safe Explorer code analysis, IBM Quantum Safe Advisor, and IBM Quantum Safe Remediator.
For more about our roadmap for the future of quantum-safe systems and tools, read our blog.
Writing software takes time and requires deep expertise that’s often hard to find. What if an AI could help software engineers get the job done faster? What if you could translate plain English to code? IBM Watson Code Assistant is powered by watsonx.ai foundation models to assist developers of all skill levels. Type your desired coding task into a source code editor like Visual Studio Code, and Watson Code Assistant will return an AI-generated recommendation. Learn more here.
Much of computing today takes place on mobile phones at the edge. But running foundation models in this environment can be slow and complicated because of data-privacy rules. Our delivery model addresses both challenges; we deploy foundation models closer to where the data is generated, to lower latency and enable near-instantaneous predictions. We are also set up to follow the latest data privacy rules.
Our watsonx.ai at the edge deployment stack allows us to create foundation models at the edge in hours instead of weeks. By pre-training models in the cloud, we can take advantage of the vast amount of public and industry data licensed by IBM. The cloud also allows us to continuously update these models for accuracy.
Learn how we’ve used our foundation model stack to carry out visual inspections at the edge at Dubendorf Air Base, the Canton of Zurich, and other enterprises.
Security analysts investigate a variety of alerts each day, many of them generated by suspect IP addresses, domain names, or file hashes. But still a lot of threats slip through. By training a foundation model on past security incidents, we can now detect warning signs that previously went unnoticed. Our watsonx.ai for security model watches a stream of network activity, looking for out-of-the-ordinary processes and behaviors that could signal a malicious attack.
There’s an old saying: work smarter, not harder. That’s the idea behind the newest version of our conversational AI platform, Watson Assistant. It’s now supercharged with watsonx.ai large-language foundation models to give customers a better experience while boosting employee productivity. And it’s not just Watson Assistant: We’re embedding new large language model technologies into all of our Watson Discovery digital labor products to help businesses improve productivity and create additional value for customers. Learn more about Watson Assistant here.
Data is the fuel that powers modern AI, but managing data from different sources with varying restrictions is one of the biggest barriers to AI adoption that enterprises face. One of the other main pillars of watsonx announced this week is watsonx.data, which simplifies this process. It’s the industry’s only data store product that can manage workloads both on-premises and across multiple clouds, at half the cost of traditional data warehouses. We also provide built-in governance and automation, as well as the ability to integrate with an organization’s proprietary data and tools, making watsonx easy to set up and use. Learn more here.
IT environments are complex and varied, making it difficult for managers to get a holistic view of the system. Foundation models for code and natural language can help. IBM has been working on ITOps automation with watsonx that can give managers of both IT operations and site reliability the ability to see across the system to resolve incidents in a fast, cost-effective way. Learn more here.