Facing the AI Challenges head-on
In just a few months, Artificial Intelligence has gone from under-the-hood technology to a go-to tool for everyone, from kids creating fantasy pictures to leaders drafting business strategies.
AI pioneer Morten Goodwin tells us why this happened and the pros and the perils of AI.
Morten Goodwin has studied and researched Artificial Intelligence (AI) technology for several years. Goodwin is a professor at the University of Agder and specializes in the technology of deep neural networks. He is also an active communicator of popular science. In 2020, he published the book: AI. Myten om maskinene (AI. The myth of the machines) at Humanist Forlag.
Since AI is on everyone's lips, the AI buzzing is incredibly high in tech businesses. It was no surprise that the ODA network invited Morten Goodwin to share his thoughts on the technology and why it’s leapfrogging at the moment. Around 300 participants from IT companies and IT departments in both public and private companies met up in the cellar at Cegal for some tapas, networking, and new AI insights.
AI works marvelously well, but we also lose some control over the technology.
Morten Goodwin.
AI is a big new opportunity for the IT industry. Programmers, for example, suddenly have a lot of tools that will help them program faster and better,” he explained and showed the audience a «Space Invaders» game developed by a journalist, with no coding skills, based on GPT prompts.
ChatGPT suggests five bullet points where AI shows the most promise:
● Automated Software Testing: AI can identify bugs and optimize code more efficiently than manual methods.
● Natural Language Processing: Tools for chatbots, translators, and sentiment analysis.
● Predictive Analytics: From sales forecasts to predicting machine failures.
● Image and Video Analysis: Automated tagging, facial recognition, and anomaly detection.
● Gaming and Simulation: Creating more realistic virtual environments and NPC behaviors.
What triggered the AI tipping point?
Every tech-savvy person knows AI technologies have been around for several decades and that the answers or outcomes may be biased or flavored. The technology isn’t better than the data it’s trained on.
So why do we have this AI tipping point right now? According to ChatGPT, there are four primary triggers:
1. Advancements in Technology: Hardware advancements like GPUs, TPUs, and cloud computing have made it feasible to run complex algorithms.
2. Data Proliferation: The surge in data available from various sources such as social media, IoT devices, and businesses.
3. Economic Factors: Decreasing costs of using and implementing AI solutions.
4. Societal Awareness and Acceptance: As AI became more prevalent in popular media and basic technologies, its societal acceptance grew.
Goodwin presented slides that emphasize the advancements in technology. When Alphabet (Google) launched Bert in 2018, it was trained on 100 million parameters. A few months later, Elon Musk released GPT-1 trained on 110 million parameters. Later, in 2018, Alphabet launched Bert Large, which trained on 350 million parameters, and GPT-2 (2019) was trained on 1.5 billion parameters. GPT-4, launched earlier this year, was trained on 100 trillion, almost 67,000 times larger than GPT-2.
There is no moral or even truth in AI
AI professor Morten Goodwin underlind that natural language processing (NLP) models do not have built-in morals or ethics. They don’t even know the truth. There are a lot of examples of ChatGPT making up referrals and references as sources for assertions in the text that don’t exist.
ChatGPT is trained to hold a conversation. Sometimes you will get the true and correct answers, but very often it can be challenging to sort out what is true and what is mixed up.
ChatGPT concludes: “The integration of AI into our daily lives is undeniable. While it brings a plethora of opportunities, we must approach it with caution and wisdom. As we stand on the cusp of this AI tipping point, let's ensure we make the best out of it without overlooking its challenges.”
