Generative artificial intelligence can provide basic designs for cable runs, conference rooms, lighting systems and more. Tim Kridel explores how other professions are using it and what pro AV can learn.
One year ago at this time, ChatGPT was unheard of. Then almost overnight, everyone everywhere was talking about it, the same way Covid-19 burst into the mainstream in early 2020. The similarities don’t end there, either. Like Covid in its first year, ChatGPT has a lot of people wondering if life will ever be the same again.
The short answer is no — but not necessarily in ominous or obvious ways. A prime example is the fear that ChatGPT and other types of artificial intelligence (AI) will steadily take jobs away from people. One way to understand how that scenario might play out in pro AV is by looking at other professions that are further ahead in implementing AI.
For example, the electrical industry is using ChatGPT and other generative AI tools for tasks such as designing cable raceways and lighting systems. Electrical also is an example of how pro AV might eventually have bespoke generative AI tools that cater to the industry’s specific requirements. The initial product from a startup called Augmenta is a tool that designs electrical raceways 70% faster than the traditional way of humans doing it manually on computers.
Electrical also has a chronic shortage of skilled workers. This is where the employment picture becomes more nuanced and less gloomy. Electrical contractors and design firms are using generative AI not because they want to reduce staff but rather because they can’t find enough qualified people in the first place. Sound familiar?
Electrical firms say that generative AI tools are capable of creating up to 30% of an initial design. By shouldering this preliminary grunt work, generative AI frees their highly skilled, highly paid employees to focus on building on that foundation and adding the kind of value that only humans can. This also can increase revenue if it means the firm now can take on more projects than its staff could handle without generative AI.
“I think [all] that puts even greater pressure on us to consider using tools like AI because we need our really good people doing what really good people do: spending time with customers and doing the value-added things,” says Julian Phillips, AVI-SPL senior vice president and XTG managing director. “What we don't want them doing is a bunch of repetitive tasks.”
If there’s a potentially dark lining to this silver cloud, it’s that generative AI reduces opportunities for rookies to cut their teeth on the basics of a profession. But at the same time, they’re spared a lot of drudgery and can go straight into more interesting work. As a result, AV firms could use generative AI to attract and retain employees, including those who want to use a tool that could be as transformational as the PC.
Early days
ChatGPT stands for Chat Generative Pre-trained Transformer, a moniker that’s a fancy way of saying the tool takes raw information such as images, audio, video, text and Revit files, applies machine learning to identify patterns in that input and then outputs something. In pro AV, that something could be the basic design for a conference room and a press release or case study about that project. It also can be used to generate content, such as for a client’s digital signage network as part of a managed service. These examples highlight how multiple departments can use generative AI — not just designers and engineers, but also sales, PR and marketing.
Many AV vendors and integrators already use AI — just not the generative type. But those still provide useful experiences for understanding where and how this new type of AI could be implemented. An example is learning how to determine which processes can be semi-automated to free up humans to focus on other tasks.
whiteMocca/Shutterstock.com
“Over the last couple of years, the AVI-SPL Symphony platform has been using systematic AI for auto-detecting issues, generating tickets and enabling technology managers to automate routines and provide insights on usage patterns,” Phillips says. “However, the use of generative AI has been viral and mostly unplanned up until now.
“On stage at our annual sales and design conference in February, I asked the audience who was using ChatGPT at work, and a quarter put their hands up. Admittedly, most of them were engineers using it for writing and checking code, writing statements of work and researching technology, but marketing is using it to support creative writing, sales for customer research and pretty much everyone else playing with it trying to figure out what works and what does not.”
Although the ChatGPT brand quickly became shorthand for generative AI, it’s not the only such tool on the market. Others include DALL•E 2, Microsoft’s Azure OpenAI service, Midjourney and Transcend, to name just a few. Considering all of the buzz that generative AI is generating, it’s a safe bet that even more startups and major IT vendors will enter the market. For AV firms, one challenge that comes with a big vendor ecosystem is determining which ones will still be around in a year or three to support their generative AI products.
Major AV vendors might develop their own generative AI tools for use by consultants and integrators. Until those bespoke products arrive, another consideration is which tools can be applied to pro AV without extensive, expensive customisation. This is another example of where it helps to look at other professions to see what they’re facing and doing.
“Adapting existing open-source or paid models is cost effective,” Boston Consulting Group says in The CEO’s Guide to the Generative AI Revolution. “In a 2022 experiment, Snorkel AI found that it cost between $1,915 and $7,418 to fine-tune a large language model (LLM) to complete a complex legal classification. Such an application could save hours of a lawyer’s time, which can cost up to $500 per hour.”
Another option is to build a tool from scratch and then train it. That’s expensive — the kind of investment that only major AV integrators or AV vendors can afford — but it could enable capabilities that give those firms a major competitive edge.
“Training a custom LLM will offer greater flexibility, but it comes with high costs and capability requirements: an estimated $1.6 million to train a 1.5-billion-parameter model with two configurations and 10 runs per configuration, according to AI21 Labs,” Boston Consulting Group says. “To put this investment in context, AI21 Labs estimated that Google spent approximately $10 million for training BERT and OpenAI spent $12 million on a single training run for GPT-3.2 (Note that it takes multiple rounds of training for a successful LLM.)
“These costs — as well as data centre, computing, and talent requirements — are significantly higher than those associated with other AI models, even when managed through a partnership. The bar to justify this investment is high, but for a truly differentiated use case, the value generated from the model could offset the cost.”
Why a companywide AI strategy is key
These kinds of big, complex decisions — along with the fact that generative AI can be used by multiple departments within an AV firm — highlight the importance of developing a companywide strategy. That requires buy-in from the top.
“Today’s focus might be on productivity gains and technical limitations, but a revolution in business-model innovation is coming,” Boston Consulting Group says. “Much as Mosaic, the world’s first free web browser, ushered in the internet era and upended the way we work and live, generative AI has the potential to disrupt nearly every industry—promising both competitive advantage and creative destruction. The implication for leaders is clear: today’s breathless activity needs to evolve into a generative AI strategy owned by the C-suite.”
Generative also can be one part of an overarching initiative to identify all of the ways to leverage all types of AI.
“The applications and business benefits of AI are endless and thereby lies the problem: focus,” says AVI-SPL’s Phillips. “The AV industry in particular has been slow to adopt advanced CRM tools, which are now using AI to improve opportunity qualification, workflow and conversion. Salesforce's Einstein GPT can spot opportunities within accounts, recommend next actions and automate communications. HubSpot's AI tool, Content Assistant, uses GPT to create copy, social media posts and automate follow-ups.
“In the back office, AVI-SPL has a process improvement department led by Jeremy Codiroli using LEAN/Six Sigma principles to identify margin leaks, operational improvements and supply chain efficiencies. Now AI tools are helping to accelerate change and scale the cost reductions. It is a good example of how AI is a tool to support and enhance existing business operations rather than a solution on its own.”
Partly because so many departments can use generative AI, data security should be a top consideration when developing and implementing it. Generative AI vendors depend on a steady diet of real-world input to train their tools so they can keep getting smarter. For AV firms, this creates the challenge of ensuring that proprietary/confidential client or company data doesn’t get used by the AI tool’s vendor.
“Clearly all data collection, storage and usage needs to comply with data privacy laws such as GDPR,” Phillips says. “You would definitely expect that all systematic AI should follow the same strict IT security and privacy controls.
“However, it is probably the uncontrolled use of generative AI that creates the greatest risk. For example, we have already seen the legal profession being exposed to the injudicious use of AI-generated case law which proved to be wrong, leading to a disbarment.”
Another key consideration is the data sources that generative AI uses. If they’re incomplete, it skews the tool’s output.
“The perpetuation and dissemination of ‘bias’ creates risk,” Phillips says. “For example, if I ask ChatGPT 4.0, ‘What are the best speakers for auditoriums?’ I receive great advice on design considerations, but also recommendations on particular manufacturers' products and not others. The advice to all [is that] even when you are not using AI, others are and that could create a competitive disadvantage for you.”
Phillips also sees parallels with the internet 30 years ago, when his employer at the time — Michael Dell — decided to provide the entire company with online access.
“He believed that the only way we could continue to innovate and stay ahead of our competition was to encourage curiosity, exploration and self-learning,” Phillips says. “He was right. Therefore, I would give the same advice to any business leader: provide free and unrestricted access to AI tools within the business, but also provide a framework, some tools and support to accelerate great ideas, but also expect some failure.
MMD Creative/Shutterstock.com
“I was speaking to the CEO of a well-known AV company at InfoComm and he shared that some of their top priorities moving forward involved the application of AI. I commented, ‘In that case, in addition to your COO, CMO and CFO, why haven't you got a Chief Artificial Intelligence Officer (CAIO)?’ Given that AI will be as influential as the advent of the internet, I would recommend that all businesses have a CAIO.”
Top image credit: MeshCube/Shutterstock.com