This is the real future of AI

Scientists all over the world are using AI-assisted technologies in work that has the potential to lead to major advances in medicine, agriculture and mathematics. Why are we not hearing about them?

Researchers at two different German science institutes have built AI-assisted genetic models to study the way that genes operate inside cells. Their work may one day yield new advances in food production, healthcare and medicine.

At Forschungszentrum Jülich and the IPK Leibniz Institute, researchers trained a machine learning model on hundreds of experimental DNA-binding datasets, teaching it to recognise the binding patterns of 46 transcription factor families at once. The model is allowing the team to study the ways that genes in the cells of a particular plant species turn on and off and to predict how certain patterns affect plant growth. Their work may eventually be used to predict which plant species produce higher crop yields and why.

Across the country at the Technical University of Munich, another team of scientists built an AI model, RegVelo, that traces the paths of genes through cells and can also simulate interventions in gene regulation.

“RegVelo is a step toward virtual cell models that will help us better understand how cells behave in differentiation contexts and how they respond to genetic perturbation,” lead researcher Fabian Theis explained. “Such hybrid models, combining data-driven AI with mechanistic biological structure, could be key to moving from description to prediction in biology. … In the long term it could “help us better understand disease-relevant cell states and identify possible starting points for new therapies.”

A quick note: The researchers mentioned above are working with complex computational models of certain cells. No actual genetic engineering or experimentation is taking place. Also, it must be said, no one took their data and research question and “threw it into Claude” (or CoPilot or ChatGPT) and waited for the result.

You would be forgiven if you had never heard anything about either of these studies–or the thousands of similar efforts that are underway all over the planet.

Bubble, bubble, toil and trouble …

We used to be able to rely on the news media to keep us informed, and also translate new scientific advancements or areas of inquiry into language the average reader could understand.

But even major news outlets has largely confined their coverage of generative artificial intelligence (AI) to uncritically repeating increasingly fantastical (and baseless) assertions from a handful of AI company CEOs.

First, ‘AI’ was going to replace half of all workers. Then, it became clear that large language models (LLMs) were not, in fact, capable of replacing most workers. And, even if they were, the real cost of using them was higher than the workers they would replace.

Now, the narrative du jour is that AI is super dangerous and also capricious – “breaking out” of its environments to hack the computer systems of its owners and competitors alike. We need to “slow down” its progress so that it doesn’t somehow (details are always lacking) lead to the extinction of humanity.

Strangely, the solutions proposed by the people making these claims never include more regulation of said technologies, nor the establishment of any kind of ethical guidelines or norms to ensure safe future development.

We just need to give them more money, they claim. Self-regulation will somehow work just fine, even though it hasn’t worked at all, so far.

Bad reputation

Thanks to the efforts of some exceptionally greedy and exceptionally mendacious tech billionaires – aided and abetted by too many corporate C suites and fawning tech journalists – AI now has a terrible reputation among the general public.

People rail against the infiltration of ‘AI slop’ in all areas of human endeavour. Teachers and academics warn that the use of LLMs is degrading our ability to read and to reason for ourselves. This morning, I read that Shopify CEO Tobias Lütke complained that employees’ “AI slop grenades” were drowning the offices in extra, unnecessary work.

Journalist and author Cory Doctorow, who coined the term enshittification, says that AI has now become synonymous in the public mind with “low quality.”

“No one ever said, “My kid’s math teacher was replaced with AI” in a happy tone of voice,” he wrote on a recent blog post. “No one ever said, “Oh, great, they replaced their customer service department with AI chatbots!” My teenager and her friends use “That’s so AI” as a shorthand for “That’s low-effort shit.”

The future is wealthy, lazy and stupid?

As far as most people know, ‘AI’ is the thing people who don’t like to write use to do all of their writing – from work emails, to college term papers, to overlong think pieces on LinkedIn.

Then, there are articles like this recent one in The Atlantic, claiming AI personal assistants – bots that, when given access to the user’s email, instant messages and credit card information, automate large swaths of one’s personal time – are the “future of AI.”

“Instinct lets people ask for help with all kinds of tasks by texting the bot directly through iMessage or WhatsApp. Lately, it has been all the rage in the tech world. Online, people are breathlessly sharing examples of what they’ve gotten the tool to do. … Noah Shinn, a 23-year-old Northeastern University dropout, launched the product to a small test group earlier this year, and his company is already worth billions of dollars.”

Predictably, some users have reported some significant hiccups after giving Instinct access to their money, personal data, and schedule.

“One user recently complained on X that Instinct lost him more than $200 after it prematurely canceled his flight. He reported that the bot texted him the following: “Straight story: while pulling up the cancel terms, the cancellation actually went through before I could show you the cost first.” When a man in Los Angeles asked Instinct to scan for restaurant reservations for an upcoming trip, the bot apparently made a reservation without his approval at a place that charged a $200 cancellation fee.”

I couldn’t help but wonder where we would be as a species if the people who had billions of dollars to spare decided to fund things like tools to improve cancer research or address drought and global water shortages, instead of an automated assistant for people with way more money than sense.

The truth is out there

While the big headlines about AI get ever more farcical, the real truth is that thoughtful, smart people from many different disciplines are using AI in ways that are far more consequential than deep-faking your Instagram feed or automating the cumbersome dinner-reservation-making process.

Some more examples:

This page on GitHub contains a list of “curated papers, articles, and blogs on data science & machine learning.”
This paper from the journal, Agronomy, on machine learning applications in agriculture.
This crowdsourced list from blogger Terence Tao of general resources for AI and mathematics.

To be clear, I am not arguing that the unregulated development and, particularly, deployment of artificial intelligence technology is without risk. Several groups of researchers, scientists and government leaders have begun to call for significantly more oversight and regulation of AI.

In June, the International Mathematical Union (IMU) released The Leiden Declaration on Artificial Intelligence and Mathematics, a cooperative statement on the challenges posed by the use of AI in mathematics research.

The nonprofit, ControlAI, has developed draft legislation in the UK to regulate AI development.

But as the general public has such a poor understanding of what exactly AI is, I fear that the popular opinion to just ‘turn it off’ and oppose it in all situations will lead to the abandonment of many of these initiatives that are important and could benefit all of humanity, rather than a select few.

The real promise of AI is a technological advance that–properly regulated and studied–can support and benefit human communities, not replace them.

To get a reliable understanding of what AI really is and can be used to do, I have found better information from independent writers and publications, particularly those in the scientific community. Here are some pieces related to my post with more detailed information.

Cory’s post on Pluralistic explains the alleged hacking incident with OpenAI and Hugging Face. Ed Zitron tackles Jacob Coxon’s claims that AI could end humanity in the next 10 years.

The Stanford Report: Scientists call for all-out, global effort to create an AI virtual cell

Pluralistic.net: LLMs are Real, AI is Fake

Where’s Your Ed At?: AI is Already in Dangerous Hands