The statement provided touches upon several complex issues related to risk management, prediction utility, machine learning/AI effectiveness, and societal perceptions of AI. Let's break down the key points and provide a counter-perspective to each:

Risk Profile Over Predictions:

The statement suggests using predictions only when the risk of error is low, which is a conservative approach to risk management. However, this approach could limit the potential benefits of predictive models in high-risk areas where they can be incredibly valuable, even if they're not perfectly accurate. For instance, in healthcare, even a small predictive edge can lead to better patient outcomes.
Pessimism about Prediction:

The assertion "You can't predict almost anything anyway" is too broad and pessimistic. In fact, many fields, like weather forecasting, stock market trends (to some extent), and consumer behavior, have seen significant improvements thanks to predictive models. Machine learning models can identify complex patterns that may not be apparent to human analysts.
Utility+Risk Distributions:

The idea here is that action should be based on a combined assessment of utility (the benefits of a correct prediction) and risk (the costs of a wrong one). This is a sound principle in decision theory, but it doesn't necessarily lead to the conclusion that ML/AI is not useful. Instead, it emphasizes the need for careful application and the weighing of benefits against potential risks.
Underestimation of ML/AI Utility:

Arguing that most of ML/AI isn't very useful overlooks the vast array of applications where AI has already made significant inroads, such as in recommendation systems, natural language processing, autonomous vehicles, and medical diagnostics. Each of these fields has seen AI perform tasks with a level of consistency and scale that humans cannot match.
Generative AI's Unique Contributions:

The statement acknowledges that generative AI (like GPT) has achieved something significant. Indeed, generative models have opened up new possibilities in creativity, content generation, and problem-solving that were previously difficult or impossible to automate.
Public Perception and Utility:

While it's true that there might be a shift in public perception as people experience the limitations of AI (like with Siri), this doesn't necessarily mean a return to the "status-quo-ante-openai." User adaptation and technology improvement tend to go hand-in-hand, with each influencing the other.
Financial and Legal Implications:

The mention of financial achievement refers to the considerable investment required to develop state-of-the-art AI. This investment is a testament to the value that companies and investors see in AI's potential. The legal aspect regarding "whitewashing copyright" is a contentious topic. While there are concerns about AI and copyright, the development of AI also stimulates discussions on how to evolve copyright laws to better fit the modern digital landscape.
Refuting the overarching sentiment of the statement:

AI and ML have shown remarkable utility across various sectors. To claim they are not very useful is to ignore the measurable impacts they have had on efficiency, discovery, and productivity. The utility of AI systems is not static; it improves as the technology advances and as more data becomes available. Moreover, AI's role in complementing human capabilities should not be underestimated.

The ability to predict is not binary (predictable vs. unpredictable), but rather a spectrum where the confidence in predictions can be quantified and managed. While there's truth in being cautious about over-reliance on AI, especially in critical decision-making scenarios, the current trajectory of AI development is towards more robust, interpretable, and reliable systems that are increasingly integrated into our decision-making processes, augmenting rather than replacing human judgment.

In conclusion, while the statement makes valid points regarding risk management and the current state of public understanding of AI, its usefulness and potential cannot be dismissed. The application of AI in society is a dynamic field with ongoing advancements, and its utility is likely to expand, rather than contract, in the future.