From: aidotengineer
This article summarizes key insights and discussions from a “Frontier Feud” event, where 100 AI Engineers were surveyed on various topics related to the AI industry [00:01:07]. The event highlighted several AI technological advancements, challenges and opportunities in AI adoption, and future trends in the field.
Emerging AI Technological Advancements and Trends
Participants shared their “tech hot takes” which touched on future trends and developments in the AI space:
- Shift in AI Model Training: A prediction was made that at least one of the current five major entities training big models today will cease to do so by the end of the year [00:01:55]. This suggests potential consolidation or strategic shifts among leading AI developers.
- On-Device and Smaller Models: It is projected that within a year and a half, the majority of deployed AI models will be on-device [00:03:34]. This trend favors smaller models that are orchestrated together and hyperspecialized for specific tasks, moving away from sole reliance on larger, cloud-based models [00:03:40]. This represents a significant future of AI technology trend.
- AI for Interpersonal Interactions: A “hot take” suggested that dating apps in the future might involve AI agents dating each other [00:04:13]. Another prediction indicated that future conversations would involve at least two other bots in the room, suggesting a more multi-agent interaction model [00:04:40].
- Biological AI: A long-term perspective suggested that Transformers might not be the final AI architecture, with future models potentially being built using biological materials [00:04:19].
Challenges and Considerations in AI
The survey questions to 100 AI Engineers revealed several key challenges and considerations within the industry.
Top Considerations When Choosing an AI Model
When selecting an AI model, the top considerations for engineers include:
- Cost/Price: This was identified as the number one consideration, which surprised many in the audience [00:10:47]. This highlights a significant enterprise AI and ROI challenge.
- Latency: The speed of the model is also a crucial factor [00:11:04].
- Accuracy/Performance: Evaluating benchmark scores and overall performance is important for model selection [00:11:16].
- Capabilities: The specific functionalities and abilities of the model are a key decision point [00:13:18].
- Open Source vs. Closed Source: The licensing model also plays a role in the decision-making process [00:12:45].
Buzzwords Engineers Are Tired of Hearing
AI Engineers expressed fatigue with certain buzzwords:
- AGI (Artificial General Intelligence): This was the most commonly cited buzzword that engineers are tired of hearing [00:15:21].
- Agents: Another frequently mentioned buzzword [00:13:56].
- RAG (Retrieval-Augmented Generation): This term is also causing fatigue among engineers [00:16:21].
- Prompt Engineering: This term was also listed as something engineers are tired of [00:16:25].
Jobs Most at Risk of AI Disruption
When asked about jobs most at risk from AI disruption, the top answers were:
- Data Entry: Identified as the number one job at risk [00:20:43].
- Software Engineers: This profession was also listed among the top three at risk [00:18:15].
- Content Creation/Writing: This broader category encompassing artists and writers was also deemed at risk [00:20:10].
Biggest Nightmares for AI Engineers
The survey revealed the biggest nightmares for AI Engineers at 2 AM:
- Model Outage: An outage of the AI model was the top nightmare [00:20:52].
- Hardware Failure/Info Problem: Issues with hardware or information were also a significant concern [00:17:55].
- Cold Email from a VC: A humorous but relatable nightmare for engineers [00:20:28].
Industry Most Benefiting from AI
Healthcare was identified as an industry that would benefit the most from AI [00:20:57].
Future of AI Engineering
A large-scale survey on the state of AI engineering is planned to gather more in-depth information on the tools and workflows used by AI Engineers, aiming to increase transparency within the industry [00:21:39]. This highlights ongoing efforts to understand and optimize AI engineering practices.