The Artificial Intelligence Show

ChatGPT Business, AI Disrupts Politics, and AI-Powered Growth and Layoffs in Big Tech

Tue May 02 2023

OpenAI Updates to Chat GPT

00:00 - 05:50

OpenAI announced two big updates to Chat GPT, including a subscription tier for enterprises and a feature that allows individual users to turn off their chat history.
The subscription tier will follow OpenAI's API data usage policy, meaning user data given to Chat GPT will not be used to train the model by default.
Individual users at any tier can now turn off their chat history, preventing their data from being explicitly used to train Chat GPT.
These changes address concerns about privacy and security when using large language models in an enterprise setting.

Concerns of Enterprises Regarding Data Usage

05:25 - 10:42

OpenAI is making a play to address the concern of enterprises not wanting their data in future training sets for the next foundational models.
Enterprises are trying to figure out which language model company or which AI application company they should be working with and how that works.
Almost every conversation at the C-suite level is coming back to building a strategy around the use of large language models.
People need to be careful what information they put into these tools, understand the usage of their data, and make sure it remains confidential proprietary data.
This will be a continuous theme on this show because almost every enterprise is starting to realize they have to solve for that and trying to figure out what do they do? Who do they talk to? How do they move forward?
OpenAI's move puts them in the conversation around corporate language AI strategy.
OpenAI has faced some pretty serious legal and regulatory challenges, notably with Italy trying to ban Chatshoot-B-Tib, and the EU looking more closely at how OpenAI trains models.
It all seems connected. OpenAI is in conversations with the US government, oversight and regulations, lobbying just kind of protect innovation side of this.

Government Regulation of AI

10:12 - 15:52

OpenAI is in conversations with the US government regarding oversight and regulations.
AI-generated attack ads are being used in the 2024 presidential election season.
Governments in the US and Europe are taking more aggressive action to regulate AI.
Four major federal agencies in the US released a joint statement on their stance towards AI companies, stating that they will not treat AI companies differently from other firms when enforcing rules and regulations.
The European Parliament reached a deal to move forward on the world's first comprehensive AI rulebook called the Artificial Intelligence Act, which governs the use of AI within the European Union.
AI is impacting governments and society, and lawmakers are reacting by taking action to regulate it.
AI can create realistic synthetic media at a level of accuracy or scale that we've never seen before, which will be used for personalization of ads, messaging, text messaging, email, video content, etc. during political campaigns.
The average person can create endless amounts of this content using widely available AI technology.
The ability to create this stuff at scale worries me because it can be targeted at an individual level.

Concerns about AI-generated Content

15:33 - 21:13

AI-generated content can be created at scale and targeted at an individual level, which is a concern for political campaigns.
Political campaigns have the ability to manipulate people's emotions and behavior like never before.
Businesses should prepare themselves to help educate society about AI-generated content.
There needs to be increased awareness that what people see online may not be real or true.
Sources of truth will become increasingly important in processing information.
The US is moving towards regulating AI by applying existing laws, while the EU is attempting to craft AI-specific legislation.
It will get hard to apply specific laws to new technologies if they are not defined.

Applying Laws to AI

20:44 - 25:52

The US is trying to apply existing laws to AI, while the EU is trying to get specific.
It's difficult to apply laws to specific technologies as new technologies keep popping up.
General guidance might be a better path than defined laws.
The joint statement from federal agencies lacks clear guidance on how they will govern AI.
Existing laws can govern obvious areas like civil rights, non-discrimination, and consumer protection.
It gets messy when looking at some civic areas like discriminatory impacts or unsubstantiated claims about AI.
FTC Act requires assessing and mitigating risks before deploying AI.
The joint statement lacks tactical guidance and examples of how this has been applied.
Automated systems that can be skewed by unrepresentative or imbalanced data sets are problematic.
Lack of transparency in model opacity makes it difficult for developers, businesses, and individuals to know if a system is fair.
Developers do not always understand or account for the context in which private and public entities will use mid-systems.
Federal government's statement on AI lacks specificity compared to Copyright Office guidance from March 16th.
AI is having a major impact on the profits, earnings, and business health of big tech companies.

Impact of AI on Big Tech Companies

25:35 - 31:34

AI is having a major impact on the profits, earnings, and business health of big tech companies.
Microsoft, Google, and Meta displayed strong or better than expected results given the economic environment due to AI.
Azure revenue was up 27% year on year for Microsoft due to AI products.
Google committed to incorporating generative AI into its products moving forward.
Meta is playing catch up to retool its infrastructure for AI but still saw an unexpected increase in sales in the past quarter.
Big tech companies face enormous pressure from shareholders to get leaner and are relying on AI to evolve their businesses and/or capture efficiencies.
Dropbox cut staff by 16% (about 500 people) due to needing a different mix of skills, particularly in AI and early stage product development.
There is growing awareness of AI's potential to drive larger efficiencies in organizations which can sometimes mean reduction of workforce.
There will be increasing pressure for bigger companies to find ways to be as efficient as possible with the help of AI tools.

Adoption of AI in SaaS Companies

31:12 - 36:33

SaaS companies are under pressure to adopt AI and those that don't may become expendable.
Product teams at SaaS companies often have no idea what they're doing with AI and don't realize the disruptive force it will have on their product roadmap.
Big companies like Google, AWS, and Microsoft are trying to find markets for their technology after working on AI for 20 years.
AWS has pre-trained AI models that were previously only accessible through developers but can now be used by anyone.
Industries dealing with shortages of workers could benefit from AI.
People who are new to the topic of AI have an opportunity to apply it to their area of expertise and understand how it will affect it.
Forming an internal AI council is a good way to build responsible AI principles, gender AI policies, figure out the impact on the organization, and be proactive about planning for the future.
Being proactive with knowledge gained about AI is the best way to achieve a positive outcome.
There are simple practical steps people with little or no experience in AI can take such as learning more about it.

Building Confidence in AI

36:03 - 41:10

Develop confidence in AI by taking an online course, reading a book, or attending a conference.
Young professionals can take the initiative to learn and get confident enough to take action.
Runway ML released their Gen 1 video to video generative AI model for mobile through the runway iOS app.
Price Waterhouse Coopers plans to invest $1 billion over the next three years specifically to transform its operations with generative AI technology.

AI's Impact on Various Industries

40:54 - 46:15

Major consulting firms are investing in Genoa AI, and TurboTax is building its own Genoa AI language model for financial management.
Every company will need workers who understand AI and can guide it, human and machine.
An AI chatbot could provide healthcare answers to patients in a way that was as accurate and empathetic as a human physician.
The AI responses were actually preferred over the human physician ones and they were rated significantly higher for both quality and empathy.
Large language models make stuff up all the time, so physicians likely still need to be in the loop at all times when it comes to this stuff.
Empathy is one of those uniquely human things, but many humans lack empathy. AI can synthesize or simulate empathy.
An add-empathy button could simulate what an empathetic human would say. This could help managers who struggle to connect with people or teach humans how to be better humans.
Replit, an AI-powered software development platform, just announced an extension to their previous series B fundraising. They are now valued at $1.16 billion.

Replit's Fundraising and Jeff Hinton's Concerns

45:52 - 51:34

Replit, an AI-powered software development platform, raised over $97 million in a series B fundraising extension and is now valued at $1.16 billion.
The funds will be used to expand their cloud services and build their lead in AI for software creation.
Replit's Ghostwriter is an AI-powered coding co-pilot.
Jeff Hinton, one of the Godfathers of modern artificial intelligence, quit Google because he wants to speak up about AI risks and fears major companies are no longer developing AI responsibly.
Hinton thinks AI is going to cause some very major issues such as the generation of fake media and job loss.
Jeff Hinton's work has contributed significantly to modern AI and deep learning.
Listeners can read Genius Makers by Cade Matts or watch his interview with Jeff Hinton on YouTube for more information on this topic.
Paul Daugherty believes that Replit will play a significant role in the explosion of innovation and entrepreneurship in the future.

Engaging with the Podcast and Marketing AI Institute

51:08 - 52:03

The hosts hope to meet more listeners at Mekon in July.
They have had recent chances to interact with podcast listeners and it's been awesome.
Listeners are encouraged to reach out to the hosts on LinkedIn.
Listeners can subscribe to The Marketing AI Show on their favorite podcast app.
Marketing AI Institute offers weekly newsletters, free monthly webinars, and online courses and professional certifications.