Ahead of Dreamforce 2024, taking place Sept. 17-19, five event speakers and leaders of the artificial intelligence industry share their thoughts on the most important priorities for the near future.
From a high level, we need something akin to the medical Hippocratic oath, which governs doctors to do no harm. It’s for others to decide whether that’s regulation or something else, but we need a framing commitment.
[time-brightcove not-tgx=”true”]I often come at things from a narrative place, and I’ve always been struck by writer Isaac Asimov’s Robot series, in which he weaves meditations around how societal principles and protections are included in the laws of robotics on an almost engineered basis. Similarly, we need someone to assert a foundational principle for all of us that AI shouldn’t do harm.
On balance, at the phase we’re in right now, I see far more benefits than any actual realized negatives. I think what’s going on in medicine alone should give people a lot of enthusiasm for the positive potential in AI. That’s the field in which I’ve seen things I think are truly astonishing, and are going to lead to real revolutions in human health and quality of life for a lot of people.
Even just AI in radiology: the capacity of AI and machine learning to just do a much, much better job than human interpretation of cancer screening. And instead of turning to treatments that have low efficacy because we’re throwing a dart at the wall, we’re starting to see the capacity of AI to create bespoke, curated, data-driven conclusions about what will benefit an individual person vs. a population.
The diagnostic potential in AI, or the interface between diagnosis and treatments that will have efficacy, combined with genetics—it just really starts to get into a world that, to me, is really positive.
But we need an ethical baseplate to do no harm. How that gets actually structured and expressed, both on an engineered, technological level and a societal, governing level, is going to be one of the really big questions and challenges of the next few decades.
For the past 20 months, generative AI and large language models (LLMs) have dominated the mindshare of leaders and driven countless innovations. However, C-suite execs and AI experts need to start looking beyond the capabilities—and limitations—of LLMs and explore the larger, more profound impact that large quantitative models (LQMs) will have on their organization and industry.
While LLMs are centered on our digital world—creating content or deriving insights from textual or visual data—LQMs drive impact on the physical world and the financial-services sector. LQMs leverage physics-based first principles to generate new products in sectors such as biopharma, chemicals, energy, automotive, and aerospace. They can also analyze large volumes of complex numerical data to optimize investment portfolios and manage risk exposure for financial companies.
With LQMs, breakthroughs that were seemingly impossible 24 months ago are now bearing fruit, transforming industries and pushing the boundaries of what is possible with AI.
Enterprises are realizing they need to implement LQMs and LLMs in order to extract maximum benefits. If CEOs focus solely on LLM-powered AI solutions for customer service, marketing, document creation, digital assistants, etc., they will likely fall behind competitors who are leveraging LQMs to transform processes, create innovative new products, or solve computationally complex problems.
Over the course of the next year, our industry needs to reset the way we talk about AI to both manage expectations of what progress looks like and bring bright, creative minds with us along the way.
This will require a collective effort to communicate our vision clearly and maintain transparency around our advancements, and it will be important to do this in a way that does not create fears or make these products out to be more than just that—products.
At Runway, we’re building significantly more advanced, accessible, and intuitive technologies and tools for our millions of creative users around the world. Our successes and future growth are driven by the strong community we’ve built through our work with artists and creatives—understanding their needs and how they approach their crafts will always be the priority.
You can see this manifested through initiatives like our annual AI Film Festival, our Gen:48 short-film competition, and our new Research and Art (RNA) community Q&A sessions.
These have all provided a platform for artists, which in turn has driven our growth and mission of empowering these artists.
I think that we should be focusing on transparency and accountability, and communicating AI’s impacts on the planet, so that both customers and members of the community can make more informed choices.
We don’t really have good ways of measuring the sustainability or the labor impact of AI. And what would be useful is to develop new ways of reflecting on how switching from one type of AI tool or approach to another changes the environmental impact.
For example, Google switched from good old-fashioned AI to generative AI summaries for web search. I think that’s where customers really want more information. They want to know: What do these AI summaries represent in terms of societal and planetary impacts? In my research, we found that switching from extractive AI to generative AI actually comes with 10 to 20 times more energy usage for the same request.
We can’t opt out of new technology—and yet we don’t know how many more computers are needed; how much more energy or water is needed; how many more data centers they have to build in order for people to be able to get these AI summaries that they didn’t really ask for in the first place.
That’s where the transparency is missing because for a lot of people, they are mindful of the climate. And so I think that companies have a responsibility to their customers to say, “This is how much more energy you’re using.”
AI has the potential to transform efficiency: it gives us the opportunity to both save people time and help create audience-specific content.
I am seeing it firsthand across several companies that I’ve been lucky to work with. For example, think about a GoFundMe campaign. If AI can help you generate your narrative in a way that makes your audience more passionate about your cause, that could be monumental for someone raising money for their neighbor.
The No. 1 angst amongst our customers at Zeck is creating infographics, charts, and graphs. Such a pain. There is not a single person in the world who likes creating charts and graphs. But Zeck AI looks at your table or data and suggests, “This may look good as a pie chart,” and creates that pie chart for you. You can choose to accept it, iterate on it, or decline it. And Zeck AI will come up with red flags as you build your narrative that you wouldn’t have thought of. Just imagine the time savings for someone who typically spends hours upon hours building everything from scratch. Now it takes minutes. Mindblowing.
I am certainly not saying that AI should replace people, but AI will definitely make everyone more efficient.
In the early 1970s, programming computers involved punching holes in cards and feeding them to room-size machines that would produce results through a line printer, often hours or even days later.
This is what computing had looked like for a long time, and it was against this backdrop that a team of 29 scientists and researchers at the famed Xerox PARC created the more intimate form of computing we know today: one with a display, a keyboard, and a mouse. This computer, called Alto, was so bewilderingly different that it necessitated a new term: interactive computing.
[time-brightcove not-tgx=”true”]Alto was viewed by some as absurdly extravagant because of its expensive components. But fast-forward 50 years, and multitrillion-dollar supply chains have sprung up to transform silica-rich sands into sophisticated, wondrous computers that live in our pockets. Interactive computing is now inextricably woven into the fabric of our lives.
Silicon Valley is again in the grip of a fervor reminiscent of the heady days of early computing. Artificial general intelligence (AGI), an umbrella term for the ability of a software system to solve any problem without specific instructions, has become a tangible revolution almost at our doorsteps.
The rapid advancements in generative AI inspire awe, and for good reason. Just as Moore’s Law charted the trajectory of personal computing and Metcalfe’s Law predicted the growth of the internet, an exponential principle underlies the development of generative AI. The scaling laws of deep learning postulate a direct correlation between the capabilities of an AI model and the scale of both the model itself and the data used to train it.
Over the past two years, the leading AI models have undergone a staggering 100-fold increase in both dimensions, with model sizes expanding from 10 billion parameters trained on 100 billion words to 1 trillion parameters trained on over 10 trillion words.
The results are evocative and useful. But the evolution of personal computing offers a salutary lesson. The trajectory from the Alto to the iPhone was a long and winding path. The development of robust operating systems, vibrant application ecosystems, and the internet itself were all crucial milestones, each of which relied on other subinventions and infrastructure: programming languages, cellular networks, data centers, and the creation of security, software, and services industries, among others.
AI benefits from much of this infrastructure, but it’s also an important departure. For instance, large language models (LLMs) excel in language comprehension and generation, but struggle with reasoning abilities, which are crucial for tackling complex, multistep tasks. Yet solving this challenge may necessitate the creation of new neural network architectures or new approaches for training and using them, and the rate at which academia and research are generating new insights suggests we are in the early innings.
The training and serving of these models, something that we at Together AI focus on, is both a computational wonder and a quagmire. The bespoke AI supercomputers, or training clusters, created mostly by Nvidia, represent the bleeding edge of silicon design. Comprising tens of thousands of high-performance processors interconnected via advanced optical networking, these systems function as a unified supercomputer. However, their operation comes at a significant cost: they consume an order of magnitude more power and generate an equivalent amount of heat compared with traditional CPUs. The consequences are far from trivial. A recent paper published by Meta, detailing the training process of the Llama 3.1 model family on a 16,000-processor cluster, revealed a striking statistic: the system was inoperable for a staggering 69% of its operational time.
As silicon technology continues to advance in accordance with Moore’s Law, innovations will be needed to optimize chip performance while minimizing energy consumption and mitigating the attendant heat generation. By 2030, data centers may undergo a radical transformation, necessitating fundamental breakthroughs in the underlying physical infrastructure of computing.
Already, AI has emerged as a geopolitically charged domain, and its strategic significance is likely to intensify, potentially becoming a key determinant of technological preeminence in the years to come. As it improves, the transformative effects of AI on the nature of work and the labor market are also poised to become an increasingly contentious societal issue.
But a lot remains to be done, and we get to shape our future with AI. We should expect a proliferation of innovative digital products and services that will captivate and empower users in the coming years. In the long run, artificial intelligence will bloom into superintelligent systems, and these will be as inextricably woven into our lives as computing has managed to become. Human societies have absorbed new disruptive technologies over millennia and remade themselves to thrive with their aid—and artificial intelligence will be no exception.
Login