The Ugly Truth About Artificial Intelligence

 

What is Artificial Intelligence?

The theory and creation of computer systems that can carry out operations that have traditionally needed human intelligence, like speech recognition, decision-making, and pattern recognition, is known as artificial intelligence (AI). Machine learning, deep learning, and natural language processing (NLP) are just a few of the technologies that fall under the broad category of artificial intelligence (AI).

Many people dispute over whether the various technologies in use today truly qualify as artificial intelligence, despite the fact that the term is frequently used to describe them. 

Artificial Intelligence examples 

In its most basic form, machine learning employs algorithms that have been trained on data sets to produce machine learning models that enable computer systems to carry out tasks like translating text between languages, recommending songs, and determining the quickest route to a destination. The following are a few of the most prevalent applications of AI today:

• Chat GPT: Responds to queries or comments by generating text using large language models (LLMs).
  

• Google Translate: Translates text between languages using deep learning algorithms.

• Netflix: Builds user-specific recommendation engines using machine learning techniques based on viewing preferences.

• Tesla: Powers their vehicles' self-driving capabilities with computer vision

What are the main established AI techniques?

1. Probabilistic reasoning

Reasoning based on probability. These methods, which are frequently referred to as machine learning, derive value from the vast amounts of data that businesses collect. Techniques for revealing unknown information contained in a vast amount of data (or dimensions) fall under this area. 

2. Computational logic

These methods, which are frequently called rule-based systems, take advantage of and expand the organization's explicit and implicit knowledge. The goal of these methods is to organize existing information, frequently in the form of rules. Although these regulations are subject to manipulation by businesspeople, the technology ensures that the rules are coherent.

3. Optimization techniques

Operations research teams have long employed optimization strategies to manage company trade-offs and optimize advantages. They accomplish this by determining the best resource combinations within a particular time frame while taking into account a number of limitations. Optimization solvers are sometimes referred to as prescriptive analytics techniques since they frequently produce actionable plans of action.

What are the main emerging AI techniques?

1. Natural language processing (NLP)

 Natural language processing (NLP) encompasses both symbolic and subsymbolic computational linguistic methods that are used to identify, parse, interpret, automatically tag, translate, and create (or summarize) natural languages.

2. Knowledge representation

The goal of capabilities like knowledge graphs and semantic networks is to make data networks and graphs easier to access and analyze more quickly. For certain kinds of issues, these strategies are typically more intuitive due to their knowledge representations. Over the past three years, the use of knowledge graph approaches has rapidly increased.

3. Agent-based computing

Although it is the least developed of the well-known AI methods, its popularity is growing rapidly. Software agents are autonomous, persistent, and goal-oriented programs that perform tasks for other programs or users. For instance, chatbots are becoming more and more common.

With current solutions, two primary groups of agent applications are frequently utilized:

• Task automation agents can be more specialized (like contract validation softbots for sales automation applications) or more general (like meeting scheduling assistants in email systems).

• Programs for autonomous objects can do tasks like automatically setting the temperature (as seen in house thermostats and automobile diagnostic systems, for example).

Artificial Intelligence: The Good, the Bad & the Ugly

Artificial Intelligence: Good vs Evil. 

The topic of AI is highly contentious. With fanatics and heretics, the fervor of those who support and oppose its spread is nearly religious in nature. The fervent supporters think AI is a game-changing technological advancement that will soon improve our quality of life and benefit all people. Even some of these believers acknowledge that it can be malevolent and turn against its creators.

The heretics claim that while AI is undoubtedly artificial, it lacks intelligence. It will never be intelligent, according to some of them. In essence, it is a statistical probability model that is capable of processing vast volumes of data from the Internet but lacks the critical cognitive ability to identify and fix its own errors.

Consider the following intelligent opinions:

(1) Regarding AI, OpenAI CEO Sam Altman stated: "This will be the greatest technology mankind has yet developed." He thinks it has the power to completely overhaul almost every business, not only those that are at risk of drastic change, like healthcare, banking, and education.

(2) Elon Musk, the CEO of Tesla, believes AI has many advantages, but there are also significant hazards that must be managed. "I'm especially concerned that these models could be used for widespread disinformation," he has stated.

(3) Gary N. Smith, Pomona College's Fletcher Jones Professor of Economics, has published a great deal about artificial intelligence. The AI Delusion is the title of his book.

According to him, AI is not intelligent and could spread false information widely on the Internet. Take his piece "Internet Pollution—If You Tell A Lie Long Enough…" from January 15, 2024, for instance. He contends that:

Unquestionably, ChatGPT, Bing, Bard, and other large language models (LLMs) are incredible. Originally designed as a new and enhanced autocomplete tool, they can compose grammatically accurate essays, have human-like conversations, and produce convincing responses to questions.

People are considered to believe lies if they are told for a long enough period. Because LLMs are not built to understand word meanings, they lack a useful method of determining if the text they input and produce is true or false. This means that in the age of the Internet, a falsehood that is repeated a lot will eventually be taken as fact by LLMs.

"This self-initiated cycle of lies is probably going to get much worse. LLMs will get more and more conditioned to repeat lies as they are exposed to a greater number of deliberate and inadvertent fabrications on the internet.

Conclusion

From narrow AI to super-intelligent AI, artificial intelligence has had a significant impact on a number of industries, including healthcare and transportation. Its capacity to forecast and analyze enormous volumes of data has created new opportunities and improved efficiency. But it also poses moral dilemmas that we must carefully handle. It is essential that we keep investigating and creating AI responsibly going forward to make sure it can be a tool for improving humankind while reducing any hazards.