‘AI’ Algorithms Aren’t People – Stop Testing Them as if They Are

So much unnecessary anthropomorphizing happening in the Machine Learning (aka Artificial Intelligence) space. From calling outright fabrications of ‘data’ ‘Hallucinations’ to claiming human emotions (“I’m sorry I couldn’t help with that….”) and giving human names to interfaces, the discussions in these areas continue to be muddied more than clarified.

When Taylor Webb played around with GPT-3 in early 2022, he was blown away by what OpenAI’s large language model appeared to be able to do. Here was a neural network trained only to predict the next word in a block of text—a jumped-up autocomplete. And yet it gave correct answers to many of the abstract problems that Webb set for it—the kind of thing you’d find in an IQ test. “I was really shocked by its ability to solve these problems,” he says. “It completely upended everything I would have predicted.”

Webb is a psychologist at the University of California, Los Angeles, who studies the different ways people and computers solve abstract problems. He was used to building neural networks that had specific reasoning capabilities bolted on. But GPT-3 seemed to have learned them for free.

Last month Webb and his colleagues published an article in Nature, in which they describe GPT-3’s ability to pass a variety of tests devised to assess the use of analogy to solve problems (known as analogical reasoning). On some of those tests GPT-3 scored better than a group of undergrads. “Analogy is central to human reasoning,” says Webb. “We think of it as being one of the major things that any kind of machine intelligence would need to demonstrate.”

What Webb’s research highlights is only the latest in a long string of remarkable tricks pulled off by large language models. For example, when OpenAI unveiled GPT-3’s successor, GPT-4, in March, the company published an eye-popping list of professional and academic assessments that it claimed its new large language model had aced, including a couple of dozen high school tests and the bar exam. OpenAI later worked with Microsoft to show that GPT-4 could pass parts of the United States Medical Licensing Examination.

And multiple researchers claim to have shown that large language models can pass tests designed to identify certain cognitive abilities in humans, from chain-of-thought reasoning (working through a problem step by step) to theory of mind (guessing what other people are thinking). 

Such results are feeding a hype machine that predicts computers will soon come for white-collar jobs, replacing teachers, journalists, lawyers and more. Geoffrey Hinton has called out GPT-4’s apparent ability to string together thoughts as one reason he is now scared of the technology he helped create

But there’s a problem: there is little agreement on what those results really mean. Some people are dazzled by what they see as glimmers of human-like intelligence; others aren’t convinced one bit.

“There are several critical issues with current evaluation techniques for large language models,” says Natalie Shapira, a computer scientist at Bar-Ilan University in Ramat Gan, Israel. “It creates the illusion that they have greater capabilities than what truly exists.”

https://www.technologyreview.com/2023/08/30/1078670/large-language-models-arent-people-lets-stop-testing-them-like-they-were

Self-Evolving Artificial Intelligence?

Teaching algorithms to create novel algorithms…

Artificial intelligence (AI) is evolving—literally. Researchers have created software that borrows concepts from Darwinian evolution, including “survival of the fittest,” to build AI programs that improve generation after generation without human input. The program replicated decades of AI research in a matter of days, and its designers think that one day, it could discover new approaches to AI.

“While most people were taking baby steps, they took a giant leap into the unknown,” says Risto Miikkulainen, a computer scientist at the University of Texas, Austin, who was not involved with the work. “This is one of those papers that could launch a lot of future research.”

Building an AI algorithm takes time. Take neural networks, a common type of machine learning used for translating languages and driving cars. These networks loosely mimic the structure of the brain and learn from training data by altering the strength of connections between artificial neurons. Smaller subcircuits of neurons carry out specific tasks—for instance spotting road signs—and researchers can spend months working out how to connect them so they work together seamlessly.

In recent years, scientists have sped up the process by automating some steps. But these programs still rely on stitching together ready-made circuits designed by humans. That means the output is still limited by engineers’ imaginations and their existing biases.

So Quoc Le, a computer scientist at Google, and colleagues developed a program called AutoML-Zero that could develop AI programs with effectively zero human input, using only basic mathematical concepts a high school student would know. “Our ultimate goal is to actually develop novel machine learning concepts that even researchers could not find,” he says.

AI Transparency and Fairness

A post on efforts to further bolster AI transparency and fairness by the AI World Society.

Learning algorithms find patterns in data they are given. However, in the processes by which the data is collected, relevant variables are defined and hypotheses are formulated that may depend on structural unfairness found in society, the paper suggests.

“Algorithms based on such data could introduce or perpetuate a variety of discriminatory biases, thereby maintaining a cycle of injustice,” the authors state. “The community within statistics and machine learning that works on issues of fairness in data analysis have taken a variety of approaches to defining fairness formally, with the aim of ultimately ensuring that learning algorithms are fair.”

The paper poses some tough questions. For instance, “Since, unsurprisingly, learning algorithms that use unfair data can lead to biased or unfair conclusions, two questions immediately suggest themselves. First, what does it mean for a world and data that comes from this world to be fair? And second, if data is indeed unfair, what adjustments must be made to learning algorithms that use this data as input to produce fairer outputs?”

Cause and effect is a challenging area of statistics; correlation does not imply causation, the experts say. Teasing out causality often involved obtaining data in a carefully controlled way. An early example is the work done by James Lindt for the Royal Navy, when scurvy among sailors was a health crisis. Lindt organized what later came to be viewed as one of the first instances of a clinical trial. He arranged 12 sailors into six pairs, and gave each pair one of six scurvy treatments thought at the time to be effective. Of the treatments, only citrus was effective. That led to citrus products being issued on all Royal Navy ships.

Whether fairness can be defined by computer scientists and engineers is an open question. “Issues of fairness and justice have occupied the ethical, legal, and political literature for centuries. While many general principles are known, such as fairness-as-proportionality, just compensation, and social equality, general definitions have proven elusive,” the paper states.

Moreover, “Indeed, a general definition may not be possible since notions of fairness are ultimately rooted in either ethical principle or ethical intuition, and both principles and intuitions may conflict.”

Mediation analysis is one approach to making algorithms more fair. Needless to say, the work is continuing.

Is an ‘AI Winter’ Coming?

A BBC post speculating on whether there is a cooling off coming for AI

The last decade was a big one for artificial intelligence but researchers in the field believe that the industry is about to enter a new phase.

Hype surrounding AI has peaked and troughed over the years as the abilities of the technology get overestimated and then re-evaluated.

The peaks are known as AI summers, and the troughs AI winters.

The 10s were arguably the hottest AI summer on record with tech giants repeatedly touting AI’s abilities.

AI pioneer Yoshua Bengio, sometimes called one of the “godfathers of AI”, told the BBC that AI’s abilities were somewhat overhyped in the 10s by certain companies with an interest in doing so.

There are signs, however, that the hype might be about to start cooling off.

AI BS

or Artificial Intelligence Bull Shitake

There are a lot of claims being made, and as this article points out, not many of them are supported by strong evidence/math.

In Rebooting AI, Ernie Davis and I made six recommendations, each geared towards how readers – and journalists – and researchers might equally assess each new result that they achieve, asking the same set of questions in a limit section in the discussion of their papers:


Stripping away the rhetoric, what does the AI system actually do? Does a “reading system” really read?


How general is the result? (Could a driving system that works in Phoenix work as well in Mumbai? Would a Rubik’s cube system work in opening bottles? How much retraining would be required?)


Is there a demo where interested readers can probe for themselves?


If AI system is allegedly better than humans, then which humans, and how much better? (A comparison is low wage workers with little incentive to do well may not truly probe the limits of human ability)


How far does succeeding at the particular task actually take us toward building genuine AI?


How robust is the system? Could it work just as well with other data sets, without massive amounts of retraining? AlphaGo works fine on a 19×19 board, but would need to be retrained to play on a rectangular board; the lack of transfer is telling.

Problems with AI Transparency

As more and more business decisions get handed over (sometime blindly) to computer algorithms (aka ‘AI’), companies are very late to the game in considering what the consequences of that delegation will yield. As a buffer against these consequences, a company may want to be more transparent about how it’s algorithms work but that is not without it’s challenges.

To start, companies attempting to utilize artificial intelligence need to recognize that there are costs associated with transparency. This is not, of course, to suggest that transparency isn’t worth achieving, simply that it also poses downsides that need to be fully understood. These costs should be incorporated into a broader risk model that governs how to engage with explainable models and the extent to which
information about the model is available to others.

Second, organizations must also recognize that security is becoming an increasing concern in the world of AI. As AI is adopted more widely, more security vulnerabilities and bugs will surely be discovered, as my colleagues and I at the Future of Privacy Forum recently argued. Indeed, security may be one of the biggest long-term barriers to the adoption of AI.