Microsoft’s implementation of OpenAI’s ChatGPT has proven to be very successful, and its unexpected adoption of this cutting-edge generative AI system caught businesses like Google and Apple off guard.
However, both Google and OpenAI are dealing with class-action lawsuits alleging copyright violations related to training their AIs based on the massive amounts of data used to train these systems. Google is actively combating the threat and developing its own generative AI solution, Bard.
It’s likely that the plaintiffs in these lawsuits are unaware of how winning might affect their careers. I’m not referring to consequences from Microsoft, Google, or other companies, but rather the possibility that their own training methods could be found to be in violation of any relevant rulings and lead to future legal action from those they learned from.
This week, let’s discuss suing generative AI companies before concluding with my Product of the Week, a brand-new HP laptop that may be ideal for you if you frequently travel for work.
Litigation Can Be Risky
I’ve unfortunately had a lot of experience in court cases. I managed my own litigation for a number of decades, spent a brief period of time as an IBM legal contract employee, and have testified as an expert witness several times. Prior to changing to a very different career path, I also received legal training.
I’ve discovered that litigation isn’t at all like it’s portrayed on television. Having different perspectives on reality when they enter the courtroom, both sides are heard before the judge or jury selects the strongest argument as the victor. The winning side, which may have been at fault, feels vindicated, while the losing side typically feels taken advantage of.
The result can have unintended and severe repercussions for the losing side that may be much worse than if they had initially left everything alone or reached a settlement without a trial. Generally speaking, appeals are unsuccessful and cost around $40,000. Before judgment, the initial trial costs may be in the thousands of dollars to over a million, and judgments themselves may be very expensive.
Therefore, before filing a lawsuit, you should consider any potential unintended consequences of both winning and losing as well as your likelihood of success. The people suing the generative AI platforms are in trouble here, in my opinion, as they are not only unlikely to succeed but also risk losing their careers if they do.
Let me elaborate.
The Training Process for Generative AI
Generative AI is trained by examining enormous amounts of data and patterns, which can then be transformed into what we call inference, which is a significantly smaller and largely federated (where the contributors to the dataset have been removed) data set that is then used as the basis for the AI to operate.
Or, to put it another way, AIs analyze digital data at such a scale that it is impossible to identify specific contributors. The result of this observation process is the creation of the AI’s brain, which is an amalgam of knowledge.
Without transparency tools, which do exist in some of the most recent Ais, it should be impossible, depending on the size of the resulting data set, to link any particular person’s behavior to their intentional or unintentional contribution to the training data.
For example, a training set of numerous comedians’ audio and video broadcasts might be necessary to learn how to be a comedian. The AI would then discover which jokes were and weren’t funny based on feedback from the audience or a training operator. After learning from its mistakes, it would create its own comedy routine without relying solely on one individual.
The issue at hand is whether the outcome violates the copyrights of anyone who unintentionally and without permission assisted in the collection of the training data set.
A problem that wasn’t expected
We don’t naturally know how to do very much, like AIs, which presents an unexpected problem. Our education comes from reading about historical or fictional events and people that were either created to entertain or make a point. We learn by watching others learn, and we also learn by reading about other people’s experiences.
We frequently pick up skills from other comics when learning a trade like stand-up comedy. Comedy is a profession where imitating others is possible. In contrast to computers, which can quickly consume knowledge from thousands of people, humans lack the mental capacity and time to learn from more than a small number of intentional or accidental mentors.
The only real difference between how AI currently learns and how people currently learn is the speed at which the learning is accomplished and the amount of training data that is observed, so if computer learning from many comedians turns out to be illegal, wouldn’t it then follow that an actual human comedian learning from a far smaller number would be violating the rights of their fellow comedic competitors as well?
If those who are suing OpenAI and Google are successful, the same case law could be applied against them, which would probably result in costly penalties.
Anyone who has trained on data that originated from the plaintiff cannot be sued because most work is generally learned through observation of others and is potentially derivative.
In other words, if these plaintiffs succeed, could not other comedians sue them due to similar training methodologies, and some comedians might then potentially be prohibited from performing for infringement should their jokes appear to have originated from other people who now also want to be compensated?
Conclusion
Now that it can learn on its own, generative AI is developing at a phenomenal and almost unbelievable rate. It was developed using extensive data sets that may have crucial information about you and your spouse. Because many of the people who unintentionally provided their data for training will start to be replaced by these systems, the training process is probably under scrutiny.
This idea of suing for a share of the outcome, however, seems ill-conceived and could have a negative effect on anyone who learns from others in the future given that all human knowledge and how it is effectively transmitted comes from someone else.
Finally, the training set has no end of life, ensuring a very limited form of digital immortality by ensuring that the knowledge amassed will continue to exist for centuries after the contributor’s passing.
I don’t think the plaintiffs in these cases will win, and if they do, the judgment might have much worse effects on how we are trained than was originally thought.
HP Dragonfly Laptop G4
The HP Folio laptop with Qualcomm technology, with its enormous 21-hour battery life, continues to be my favorite laptop of all time. My heart broke when HP released an Intel-based, business-oriented Folio with a battery life of only about six hours. Though it is also Intel-based, HP’s new Dragonfly G4 has a battery life of about 13 hours, which is much better.
The HP Dragonfly G4, which costs around $1,300 for a base configuration, offers adequate but not exceptional performance for a premium-class notebook. This laptop is designed for business use and includes Intel’s vPro solution to guarantee corporate standards are met.
This laptop has a very nice fit and finish, one of the best keyboards I’ve used so far, a good webcam, powerful speakers, and surprisingly accurate color reproduction on the display. The “auto-camera-control” and “auto-keystone” functions of the Dragonfly G4 are two special features. “.
The first enables you to stream from multiple cameras at once, displaying your face along with whatever the other camera sees on the same image. The second camera can view objects at an angle while making them appear to be straight overhead thanks to the other.
These two features are a godsend considering how many people find it difficult to display content during video conference meetings. I find it surprising that more PC OEMs haven’t taken similar action.
HP’s Dragonfly G4 is an ultra-light laptop that won’t make you feel like Quasimodo when you’re carrying it in your backpack because it weighs only 2 pounds and a half. Additionally, it offers several optional screens, WAN options, and Intel processors.
The HP Dragonfly G4 laptop is not ugly at all, even though I still prefer the HP Folio laptops’ feel and appearance. I’ve chosen it as my Product of the Week due to all of its upgrades and longer battery life.