Startups

The industrial data revolution: What founders got wrong

Comment

3D illustration Rendering of binary code pattern Abstract background.Futuristic Particles for business,Science and technology background
Image Credits: MR Cole Photographer (opens in a new window) / Getty Images

Joe Hellerstein

Contributor

Joe Hellerstein is co-founder and chief strategy officer of Trifacta and the Jim Gray Chair of Computer Science at UC Berkeley.

In February 2010, The Economist published a report called “Data, data everywhere.” Little did we know then just how simple the data landscape actually was. That is, comparatively speaking, when you consider the data realities we’re facing as we look to 2022.

In that Economist report, I spoke about society entering an “Industrial Revolution of Data,” which kicked off with the excitement around Big Data and continues into our current era of data-driven AI. Many in the field expected this revolution to bring standardization, with more signal and less noise. Instead, we have more noise, but a more powerful signal. That is to say, we have harder data problems with bigger potential business outcomes.

And, we’ve also seen big advances in artificial intelligence. What does that mean for our data world now? Let’s take a look back at where we were.

At the time of that Economist article, I was on leave from UC Berkeley to run a lab for Intel Research in collaboration with the campus. We were focused all the way back then on what we now call the Internet of Things (IoT).

At that time, we were talking about networks of tiny interconnected sensors being embedded in everything — buildings, nature, the paint in the walls. The vision was that we could measure the physical world and capture its reality as data, and we were exploring theories and building devices and systems toward that vision.

We were looking forward. But at that time, most of the popular excitement about data revolved around the rise of the web and search engines. Everybody was talking about the accessibility of masses of digital information in the form of “documents” — human-generated content intended for human consumption.

What we saw over the horizon was an even bigger wave of machine-generated data. That’s one aspect of what I meant by the “industrialization of data” — since data would be stamped out by machines, the volume would go up enormously. And that certainly happened.

The second aspect of the “Industrial Revolution of Data” that I expected was the emergence of standardization. Simply put, if machines are generating things, they’ll generate things in the same form every time, so we should have a much easier time understanding and combining data from myriad sources.

The precedents for standardization were in the classical Industrial Revolution, where there was an incentive for all parties to standardize on shared resources like transportation and shipping as well as on product specifications. It seemed like that should hold for the new Industrial Revolution of Data as well, and economics and other forces would drive standardization of data.

That did not happen at all.

In fact, the opposite happened. We got an enormous increase in “data exhaust” — byproducts of exponentially growing computation in the form of log files — but only a modest increase in standardized data.

And so, instead of having uniform, machine-oriented data, we got a massive increase in the variety of data and data types and a decrease in data governance.

In addition to data exhaust and machine-generated data, we started to have adversarial uses of data. This occurred because the people involved with data had many different incentives for its use.

Consider social media data and the recent conversations around “fake news.” The early 21st century has been a giant experiment in what makes digital information viral, not only for individuals but for brands or political interests looking to reach the masses.

Today, much of that content is in fact machine-generated, but it’s machine-generated for human consumption and human behavioral patterns. This is in contrast to the wide-eyed “by people, for people” web of years ago.

In short, today’s data production industry is incredibly high volume, but it is not tuned for standard data representations, not in the sense I expected at the time of those predictions over a decade ago.

The state of innovation: AI versus human input

One thing that has clearly advanced substantially in the past decade or so is artificial intelligence. This sheer volume of data we are able to access, process and feed into models has changed AI from science fiction into reality in a few short years.

But AI is not as helpful in the business data processing domain as we might expect — at least not yet. There is still a surprising disconnect between AI technology like natural language processing and structured data. Even though we’ve had some progress, for the most part, you can’t talk to your data and expect much back. There are some situations where you can Google for a quantitative question and get back a little table or chart, but that’s only if you ask just the right questions.

For the most part, AI advances are still pretty divorced from stuff like spreadsheets and log files and all these other more quantitative, structured data — including IoT data. It turns out the traditional kinds of data, the kinds of data we’ve always put in databases, has been much harder to crack with AI than consumer applications like image search or simple natural language question answering.

Case in point: I encourage you to try asking Alexa or Siri to clean your data! It’s funny, but not very helpful.

Popular applications of AI haven’t projected back yet to the traditional data industry, but it’s not for lack of trying. Lots of smart people at both universities and companies haven’t been able to crack the nut of traditional record-oriented data integration problems.

Yet, full automation evades the industry. Part of that is because it’s hard for humans to specify what they want out of data upfront. If you could actually say, “Here’s precisely what I’d like you to do with these 700 tables,” and follow up with clear goals, maybe an algorithm could do the task for you. But that’s not actually what happens. Instead, people see 700 tables, wonder what’s in there and start poking around. Only after a lot of poking do they have any clue what they might want to happen to those tables.

The poking around remains creative work because the space of ways to use the data is just so big and the metrics of what success looks like are so varied. You can’t just give the data to optimization algorithms to find the best choice of outcome.

Rather than waiting for full automation from AI, humans should get as much help as they can from AI, but actually retain some agency and identify what is or isn’t useful, then steer the next steps in a certain direction. That requires visualization and a bunch of feedback from the AI.

Understanding the impact of data and controlling data spread

One place AI has really shined, though, is in content recommendation. It turns out that computers are frighteningly effective at targeting and disseminating content. And oh boy, did we underestimate the incentives and impacts around that aspect of data and AI.

Back then, the ethical concerns we had around data and its uses in AI were mostly around privacy. I remember big debates about whether the public library should have digital records of the books you reserve. Similarly, there were controversies over grocery loyalty card programs. Shoppers didn’t want grocery chains to keep track of what food they bought when and target them for accompanying items.

That mentality has largely changed. Today, teenagers share more radically more personal information on social media than the brand of food they purchase.

While I wouldn’t say that digital privacy is in a good state, it is arguably not the worst of our data problems today. There are issues such as state-funded actors trying to introduce mayhem into our social discourse — using data. Twenty years ago, very few people saw this stuff coming our way. I don’t think there was a great sense of the ethical questions of what could go wrong.

This leads to what’s next, and even currently in process, in the evolution of our uses of data. What becomes the role of governments and of well-meaning legislation? Without predicting all the ways tools will be used, it’s hard to know how to govern and restrict them intelligently. Today, we are in a state where it seems like we need to figure out the controls or incentives around data and the way it is promulgated, but the tech is shifting faster than society is able to figure out risks and protections. It’s unsettling, to say the least.

So, were the predictions spot-on?

As a professor, I’d award it a passing grade, but not an A. There is substantially more data available to us with more uses than we probably ever could have imagined. That’s led to incredible advances in AI and machine learning along with analytics, but on many tasks, we’re still just scratching the surface, while on others we’re reaping the whirlwind. I am fascinated to see what the next 10 to 20 years will bring and look back on these issues again.

More TechCrunch

Intuitive Machines made history when it became the first private company to land a spacecraft on the moon, so it makes sense to adapt that tech for Mars.

Intuitive Machines wants to help NASA return samples from Mars

As Google revamps itself for the AI era, offering AI overviews within its search results, the company is introducing a new way to filter for just text-based links. With the…

Google adds ‘Web’ search filter for showing old-school text links as AI rolls out

Ilya Sutskever, OpenAI’s longtime chief scientist and one of its co-founders, has left the company. OpenAI CEO Sam Altman announced the news in a post on X Tuesday evening. pic.twitter.com/qyPMIcvcsY…

Ilya Sutskever, OpenAI co-founder and longtime chief scientist, departs

Blue Origin’s New Shepard rocket will take a crew to suborbital space for the first time in nearly two years later this month, the company announced on Tuesday.  The NS-25…

Blue Origin to resume crewed New Shepard launches on May 19

This will enable developers to use the on-device model to power their own AI features.

Google is building its Gemini Nano AI model into Chrome on the desktop

It ran 110 minutes, but Google managed to reference AI a whopping 121 times during Google I/O 2024 (by its own count). CEO Sundar Pichai referenced the figure to wrap…

Google mentioned ‘AI’ 120+ times during its I/O keynote

Firebase Genkit is an open source framework that enables developers to quickly build AI into new and existing applications.

Google launches Firebase Genkit, a new open source framework for building AI-powered apps

In the coming months, Google says it will open up the Gemini Nano model to more developers.

Patreon and Grammarly are already experimenting with Gemini Nano, says Google

As part of the update, Reddit also launched a dedicated AMA tab within the web post composer.

Reddit introduces new tools for ‘Ask Me Anything,’ its Q&A feature

Here are quick hits of the biggest news from the keynote as they are announced.

Google I/O 2024: Here’s everything Google just announced

LearnLM is already powering features across Google products, including in YouTube, Google’s Gemini apps, Google Search and Google Classroom.

LearnLM is Google’s new family of AI models for education

The official launch comes almost a year after YouTube began experimenting with AI-generated quizzes on its mobile app. 

Google is bringing AI-generated quizzes to academic videos on YouTube

Around 550 employees across autonomous vehicle company Motional have been laid off, according to information taken from WARN notice filings and sources at the company.  Earlier this week, TechCrunch reported…

Motional cut about 550 employees, around 40%, in recent restructuring, sources say

The keynote kicks off at 10 a.m. PT on Tuesday and will offer glimpses into the latest versions of Android, Wear OS and Android TV.

Google I/O 2024: Watch all of the AI, Android reveals

Google Play has a new discovery feature for apps, new ways to acquire users, updates to Play Points, and other enhancements to developer-facing tools.

Google Play preps a new full-screen app discovery feature and adds more developer tools

Soon, Android users will be able to drag and drop AI-generated images directly into their Gmail, Google Messages and other apps.

Gemini on Android becomes more capable and works with Gmail, Messages, YouTube and more

Veo can capture different visual and cinematic styles, including shots of landscapes and timelapses, and make edits and adjustments to already-generated footage.

Google Veo, a serious swing at AI-generated video, debuts at Google I/O 2024

In addition to the body of the emails themselves, the feature will also be able to analyze attachments, like PDFs.

Gemini comes to Gmail to summarize, draft emails, and more

The summaries are created based on Gemini’s analysis of insights from Google Maps’ community of more than 300 million contributors.

Google is bringing Gemini capabilities to Google Maps Platform

Google says that over 100,000 developers already tried the service.

Project IDX, Google’s next-gen IDE, is now in open beta

The system effectively listens for “conversation patterns commonly associated with scams” in-real time. 

Google will use Gemini to detect scams during calls

The standard Gemma models were only available in 2 billion and 7 billion parameter versions, making this quite a step up.

Google announces Gemma 2, a 27B-parameter version of its open model, launching in June

This is a great example of a company using generative AI to open its software to more users.

Google TalkBack will use Gemini to describe images for blind people

Google’s Circle to Search feature will now be able to solve more complex problems across psychics and math word problems. 

Circle to Search is now a better homework helper

People can now search using a video they upload combined with a text query to get an AI overview of the answers they need.

Google experiments with using video to search, thanks to Gemini AI

A search results page based on generative AI as its ranking mechanism will have wide-reaching consequences for online publishers.

Google will soon start using GenAI to organize some search results pages

Google has built a custom Gemini model for search to combine real-time information, Google’s ranking, long context and multimodal features.

Google is adding more AI to its search results

At its Google I/O developer conference, Google on Tuesday announced the next generation of its Tensor Processing Units (TPU) AI chips.

Google’s next-gen TPUs promise a 4.7x performance boost

Google is upgrading Gemini, its AI-powered chatbot, with features aimed at making the experience more ambient and contextually useful.

Google’s Gemini updates: How Project Astra is powering some of I/O’s big reveals

Veo can generate few-seconds-long 1080p video clips given a text prompt.

Google’s image-generating AI gets an upgrade