IEX #220: Why You Need a Robotics Strategy Now.
2024 could well be a break out year for robotics adoption. Do you have a strategic response ready?
Good morning, even as we go crazy about Generative AI, it’s usful to remember that 15 months ago, we were all about the Metaverse, and a year before that everything was going to be Web3.0. Technology has a way of surprising us, and the next surprise might be around the corner.
Do You Need a Robotics Strategy?
This might sound like a strange question - especially if you’re not actually using or thinking about robots in your business. Or using them tactically in warehouses or back office operations. Or maybe you think that robotics is something for the future. But what if I told you that the smartest thing you could be doing is setting up a focused effort on robotics right now. Call it a Centre of Excellence (CoE) or a Strategic Unit, or whatever you’d like. But I’m going to lay out an argument here that you need to have some kind of strategic effort in place today.
My primary hypothesis here is that we are on the verge of a breakthrough product in robotics, which will do to the space what ChatGPT did to AI. Bring it to everybody’s consciousness and give it worldwide appeal. It’s quite likely it won’t be quite the same simply because unlike ChatGPT, robots have a physicality, which means you can’t just sit in your living room with your laptop and access the service. You’d have to buy or rent a robot to use it. But that said, from a technology maturity perspective, we should be seeing a wow moment for Robotics. Perhaps a better analogy is AlphaGo or the early GPT systems which showed us the art of the possible.
Let’s take a closer look at what’s been happening in robotics.
Humanoid Robots
You often see the headlines being grabbed by Boston Dynamics (Atlas) or Hanson Robotics (Sophia) for their humanoid robots, but there are plenty of others. We are now getting to a stage where robots can mimic human motion, dexterity, and thanks to Gen AI, even language, and facial expressions. Some of the features you’ll see here include better actuators for improved dexterity, carbon fibre other similar composite metals which are lighter as well as stronger, self balancing bipedal locomotion, and structures that mimic human musculoskeletal systems. These will be coupled with high-res vision sensors, tactile sensing, machine learning for perception and decision making, and computer vision for recognising people and objects. In short, getting ever closer to human ability.
There is an underlying assumption here that the human form and traits are unequivocally desirable for robots, which is arguable. But no doubt we’ll see more and more human like robots, which if nothing else will pave the way for capabilities beyond human quickly enough. Soft robotics, which uses flexible and compliant materials which can mimic biological forms may explore other biological entities or even imagined ones.
But there’s more. And this is where we need to get beyond human abilities or limits. Some of these are inspired by nature, and some are inspired by machines, or just by the future possibilities. Some examples:
Swarm Robotics: a large number of simple robots to work together as a swarm, such as ants and bees, to be used in search and rescue, environmental monitoring, and construction, for example.
Robotic Surgery: Surgical robots like the da Vinci System are transforming minimally invasive procedures, allowing for greater precision, flexibility, and control during complex operations - in ways that are definitely beyond human capability.
In addition to all of this, there is a significant area of robot human cooperation that includes cobots, exoskeletons, drones, and autonomous vehicles. These are all areas where machines have to work in close proximity to, with, and around humans, without harm or injury. We are also on the verge of biohybrid robots which will merge robotics with muscle tissues, bacterial cells, and even living organisms, which would create truly ‘bionic’ beings.
In all of this remember that we’re also getting learning behaviours into robots - including reinforcement learning, imitation learning, and transfer learning.
The bit where this gets truly science-fictiony is the idea of phase changing robots. These are truly shape shifting entities which are also able to follow instructions and perform tasks.
Health and Elderly Care as a Special Case
Healthcare and elderly care is a particularly interesting area. We could see robotics being used as assistive tech, rehabilitation tools, social companions, telepresence support, pharmacists, hospital maintenance systems, and even nursing.
Banking and Financial Services
It’s an interesting question as to what the use of robotics might be in an area such as banking or insurance. Autonomous robots and drones will undoubtedly become a part of insurance companies toolkits for damage and claims assessment. One of the interesting areas here is the evolution of intelligent ATMs which can integrate stronger security via face and voice recognition.
Narrow Intelligence and Smart Machines
Which begs the question about when do smart machines become robots? A question we might all ask in a number of different areas. When does your smart bureau become a robot? When it gets an AI brain? Or eyes? Or can move around your room depending on your needs? When it can manage your wardrobe and fold your clothes? Could we think about narrow intelligence in robots the way we do in AI - i.e. robots/ machines that specialize in a specific area of work? Or would this be economically unviable in the short term given the investments required?
Regulatory Scramble Incoming
As we see governments across the world today scrambling to find the right governance framework for AI, it’s likely that a similar and sudden urgency will arrive should robotics have these break through capabilities. The speed of evolution creates the regulatory vaccum where guidelines and oversight are not available, but the technology is being deployed rapidly. With robots, the physical interaction and autonomous movement are additional areas of concern. Physical safety will be one of the most critical areas, along with the usual challenges of data privacy, inequality, and displacement. No doubt, there will be a wide variety of governmental positions, so the global technology evolution will see a lot of local regulatory regimes, each of which will reflect the values and challenges of their respective countries. The lack of global standards will play into this. There is a circularity here, as poor regulatory responses may stymie broader adoption if it causes harm and / or erodes trust.
Robotics Strategy
Hopefully all of this makes it clear why I say that you should start thinking seriously about robotics. Without a strategy, you might find yourself acquiring technology in an ad hoc manner, and creating potential downstream challenges with integration, costs, security, and robotic technical debt. So what should a strategy include? Think of it as a short, contained but clearly articulated document which covers the following areas:
Vision and Objectives: laying out the specific ways in which robotics will help you support and enhance your business strategy.
Use Case Identification: breaking down the directions into specific areas / starting use cases.
Creating a Technology Navigator: exploring the current and emerging trends in tech across hardware, software, AI/ML etc. to constantly track the space, and create your own roadmap through the navigator.
Evaluation of providers - build a list of potential providers and create appropriate evaluation models.
Identifying your key areas and initiating pilots: with scale-up plans and integration with existing systems to follow.
Establish lifecycle models - including programmability, extensibility, interoperability. Also consider energy requirements, maintenance costs, training, and key skill acquisition.
Infrastructure and IT Readiness: assess existing IT infrastructure (facilities, network, computing power, cybersecurity) for robot deployment, and create protocols for data management, interoperability and system resilience.
Skills and Workforce Transformation: Identify skills gaps and plan for workforce reskilling/training programs, and identify approaches for human-robot collaboration and change management
Risk Mitigation and Governance: Develop risk assessment frameworks covering safety, ethics, privacy and compliance aspects, and establish clear policies, standards and governance models for robot deployment and use
Financial Modeling: Build models for robotics cost-benefit analyses, ROI projections and develop business cases specific to your business, and explore financing options like leasing models for capital-intensive deployments
Ecosystem Partnerships: Identify strategic partnerships with researchers, user experience specialists, robot manufacturers, integrators, and your existing partners. Ensure knowledge sharing within your existing ecosystem.
Continuous Improvement: ensure key metrics capture, and drive measurement and improvement methods. Plan for continuous process optimizations and scaling of robotic capabilities
Ensure that your IT, HR, Finance and admin teams are aware and participating in your robotics strategy, so they can prepare their own policies and plans. Change management and impact on work and jobs should be key. As should the evolution of safety protocols.
Review constantly. Change is likely to be frequent.
BTW, I used Claude (Anthropic), Gemini (Google) and ChatGPT (Open AI) for researching this. I found them to be slightly different from each other. Particularly ChatGPT kept asking me contextual questions as it was answering mine so it naturally led to more sophisticated conversations.
AI Reading
Chip Redesign: The constraints of the current computing models is clearly one of the challenges in scaling AI. One of the companies taking this on is MatX, which is looking to redesign silicon based computing to optimize for LLM performance. (MatX/ Bloomberg)
Anthropic: Amazon doubles down on its Anthropic investment, adding $2.75 bn to bring its total investment up to $4bn. (Amazon)
Top 100 AI: Andreessen Horrowitz’s report on the state of Gen AI and its top 100 lists. (A16Z)
Decline of inflection AI: Microsoft’s ‘non-acquisition’ of Inflection AI and its leadership team - what does it say about the challenges of building and scaling a Gen AI product? The technical detail is in the limitations of the ‘context window’. (Fast Company/ Medium)
The Brooks Institute: report on AI in healthcare points to the challenges of data scarcity in wide scale diagnostics, the risk of monopolies, and recommends the setting up of Health Information Exchanges. (Brooks Institute)
Music making algorithms: from Meta to Stability AI, everybody’s looking at AI that can make music. But can true art emerge without pain?
Prompt Engineering: Using the question as a prompt. This piece argues that getting Gen AI tools to tell you what questions to ask is a clever prompting tactic.
Other Reading
SBF Jail Term: Sam Bankman Fried was sentenced to 25 years in Jail, in part because he had always played fast and loose with regulations, but bitcoin and cryptocurrencies have continued to flourish.
Linda Doyle: is the first female provost of Trinity College, Dublin, and wants to combine creative arts with engineering.
Forgetting & The Brain: I loved this piece about why we forget and how it’s connected to ageing.
Middle-Aged Founders: in a reassuring reminder for middle aged founders, the story of Morris Chang is told here. He started TSMC, the worlds largest semiconductor manufacturing company at the age of 55. Data suggests that middle aged founders unsurprisingly bring the wisdom that their 20 something colleagues sometimes lack. (WSJ)
thanks for reading, see you next week.