Texas House passes bill to create AI advisory council

A recent bill that passed the House would establish an advisory council to monitor the rise and adoption of artificial intelligence (AI), which has become an increasing concern for present and future generations of Texans.

Rep. Giovanni Capriglione introduced House Bill (HB) 2060 in an effort to “study and monitor artificial intelligence systems developed, employed, or procured by state agencies.”

The council would include seven members: one from each legislative chamber, an executive director, and four members appointed by the governor. Those four appointed members will include an ethics professor, an AI system professor, an expert in law enforcement, and an expert in constitutional and legal rights.

Additionally, the state council will produce a report on whether an AI system has been used in a state capacity as an automated final decision system “that makes final decisions, judgments, or conclusions without human intervention,” or an automated support decision system that “provides information to inform the final decision, judgment, or conclusion of a human decision maker.”

The State Bar of Texas has provided insight into how AI is changing the way law practices and legal judgments are being decided, including how “the use of AI for predictive analytics remains one of the biggest attractions for lawyers and their clients.”

“Even proponents of AI’s use in analyzing vast amounts of data and predicting such things as likely verdict ranges or a judge’s predispositions based on past rulings agree that technology has its limitations.”

The recent phenomenon of ChatGPT has taken hold of the cultural and political consciousness for its potential as a disruptive technology. ChatGPT is a large language model (LLM) that was developed by OpenAI to create human-like conversations through an AI chatbot.

OpenAI’s founder Sam Altman recently said in an interview that “we are a little bit scared” of ChatGPT’s success and that more regulation is important to deter its possible downsides.

LLMs use a “neural network” of informational data to create a probabilistic model of patterned language. This means that when someone asks a question in ChatGPT, the AI model creates a coherent response one word at a time, based on the overall probability that the next word in the sentence is “correct.”

Neural networks are an adaptive method of AI learning, modeled on neurons in the human brain, that teach a computer system to process data. These systems create relationship models between information from inputs and outputs of data, utilizing a technique called “deep learning” to take unstructured information from inputs and make models of probable outputs.

ChatGPT, an LLM that utilizes a neural network, does not create new information but rather “intuits” it. It is thus not “true AI” because it does not “think” as human beings do, but instead uses algorithmic predictions to mimic human intelligence in its responses.

Another task of the council is to review “the effect of the automated decision systems on the constitutional or legal rights, duties, or privileges of the residents of this state.” This relates to “AI alignment,” or the process of making sure an AI does what its human creators intend it to do.

In a recent interview with Tucker Carlson, Elon Musk sounded the alarm about what could happen with a “misaligned” AI system.

“AI is more dangerous than, say, mismanaged aircraft design or production maintenance or bad car production,” Musk said, “in the sense that it … has the potential of civilization destruction.”

An important aspect of the Texas AI advisory council will be to assess the biases that might be present in AI systems. AI bias is a well-documented occurrence; for example, the Manhattan Institute notes, “OpenAI’s content moderation system is more permissive of hateful comments made about conservatives than the exact same comments made about liberals.”

The European Union has also expressed concerns related to AI and instituted its own research commission to study its potential benefits and pitfalls.

Despite warnings from those like Musk and Altman, independent creators and AI developers have been deploying OpenAI to create a plethora of unique tools. Everything from text-to-speech vocalization and photo editing to relationship matchmaking and recipe and meal plan creations can utilize AI, which is still in just the beginning stage of what is possible with the technology.