Technology adoption of generative AI has the potential to be more significant than the advent of the internet. The majority of creative endeavors are being stifled by it, and it is far from its potential as it will be by the end of the decade.
Gen AI will make us reevaluate how we collaborate, create, solve problems, govern, and even how and whether we travel.The list provided here is by no means exhaustive.I anticipate that once this technology reaches maturity, the list of things that have not changed will be much shorter than the list of things that have.
This week, I’d like to concentrate on three issues that we should start talking about because they represent some of the more significant risks associated with generative AI. I’m not against technology, and I’m not stupid enough to suggest that it be put on hold because doing so would be impossible right now.
I propose that we start to think about mitigating these issues before they cause significant harm. Data center load, security, and relationship deterioration are the three issues.
I’ll wrap up with my Product of the Week, which is possibly the best electric SUV currently on the market. But more on that later. I’m suddenly in the market for a new electric car.
Loading of the data center
Despite all the hype, generative AI isn’t even being used by many people to its fullest capacity. It will not be possible to have the technology only exist in the cloud due to the size, cost, and resulting latency being unaffordable. The technology is processor- and data-intensive while being very personally focused.
The best strategy will probably be a hybrid where the processing power is kept close to the user, similar to what we have done with other data and performance-focused applications. To safeguard the constrained storage capacities of the client devices, smartphones, and pcs, the massive data, which will require aggressive updating, will need to be more centrally loaded and accessed.
However, because we are discussing a system that is becoming more intelligent, it may occasionally need to have very low latency, such as when it is used for conversations, translation, or gaming. It is likely that the success of a particular implementation will depend on how the load is divided without impairing performance.
While wireless technology has advanced, it can still be unreliable due to weather, tower placement, user positioning, maintenance interruptions, man-made or natural disasters, and less than full global coverage. Achieving low latency won’t be simple. While limiting data traffic and averting disastrous outages, the AI must function both online and offline.
Even if we were able to centralize everything, the cost would be prohibitive, though we do have underutilized performance in our personal devices that could reduce that cost significantly. One of the first businesses to recognize this as a problem is Qualcomm, which is working very hard to find a solution. Given how quickly generative AI is developing and how relatively slowly technology of this kind is developed and introduced to the market, it is still likely to be too little, too late.
Security
I was a competitive analyst trained in ethical ways to breach security as well as an internal auditor with a focus on security. I discovered that if someone can gather enough data, they can more precisely estimate the data they don’t have access to.
For example, you can reasonably estimate the number of employees a company has if you know the typical number of vehicles in the parking lot. The company’s top employees’ interests can typically be found on social media, and you can look at job openings to see what kinds of potential future products the business is probably working on.
These large language models gather a ton of data, and I anticipate that many of the items they input should or are confidential. Additionally, if enough data is gathered, it will become easier to fill in the gaps left by things that are not scanned in.
This scenario doesn’t just apply to corporate data. We’ll be able to learn a lot more about users’ personal lives with the kind of readily accessible personal information.
Whistleblowers, disgruntled or unfaithful workers, poor employee conduct, and workers abusing the company will all be easier for employers to identify. Protecting yourself from a hostile party discovering private information about you, your business, or even your government is becoming more realistic and accurate than I ever experienced as an auditor or competitive analyst.
The best defense is probably to spread enough false information that the tools don’t know what is real and what isn’t. This course of action will, however, significantly reduce the overall dependability of the connected AI systems, which is acceptable if only the rival uses them. However, it is likely to compromise the company’s systems, leading to an increase in unwise choices and potentially compromising the company’s systems that it wants to protect.
Interpersonal Connections
Businesses such as Suki, which provides employee supplementing avatars, and Mindverse, which offers MindOS, are demonstrating the potential personal applications of generative AI as a tool that can exhibit itself as an individual. The more we use tools like this, the less discernible it will be for us to distinguish between what is real and what is digital, and the more we focus on the tool rather than the user when judging those who use it.
Imagine having your digital twin conduct a virtual interview, represent you on dating apps, or handle the majority of your everyday virtual communications. The tool will make an effort to respond to whoever is interacting with it, it won’t grow weary or irritable, and it will be programmed to show you in the best possible light. But as it moves along this path, it will become less and less like you, and it will probably end up being far more fascinating, alluring, and composed than you could ever be.
This will be problematic because, similar to actors dating someone who has fallen in love with a role they have played, the reality will lead to more breakups and a deterioration of trust.
The simplest way to fix this would be to either learn how to act like your avatar or use them in conversations with friends and coworkers. Although I doubt we’ll succeed either, these are the two most practical ways to lessen the impact of this impending issue.
To sum up
When generative AI becomes more widely available and reaches a critical mass of users, it will greatly enhance performance. It is an incredible technology. However, a number of important issues remain that will require attention, such as overcrowding in data centers, which should propel hybrid solutions in the future, the incapacity to stop stealing secrets from these massive language models, and a notable decline in interpersonal trust.
It should be possible to prevent these hazards by being aware of them. We may come to regret some of the unintended consequences of using this technology, though, as the fixes aren’t very good.
Ocean Fisker
A towing mishap last month destroyed the battery pack in my Jaguar I-Pace, causing it to finally die. The end result was an estimate of $100,000 to fix the car, which is currently more like $40,000. My insurance company, USAA, is likely to declare the car totaled. I’ve been researching replacement electric vehicles, and the availability is terrible everywhere.
The main reason I’m probably going to buy another Jaguar I-Pace is that I don’t want to have to wait months or even years to own a car. I currently share my wife’s Volve XC-60, and we frequently have conflicting schedules that require us to use the car at the same time. When I went looking for a new electric SUV, the Fisker Ocean was the greatest model I could find.
I can’t afford to wait months for a new electric, as is the case with most of them. The Fisker Ocean fulfilled every requirement among the electric vehicles that are offered this year. Among its characteristics are:.
range of 350 miles (the bar is 300 miles).
One possible way to access the car is through your smartphone.
Use reverse charging to power your home in the event of a power outage with your car.
A remarkable time of roughly 3:06 from start to finish (I enjoy performance).
a roof covered in solar panels to extend range and provide backup electricity.
a mode that resembles a convertible, which truly opens up the vehicle.
among the most minimalistic designs available.
The Fisker Ocean is a very remarkable vehicle. I’d place my order for one right away if I could wait until the end of the year. Unfortunately, it’s not true. The Fisker Ocean is still my pick for Product of the Week, though.